Changes in Albuminuria Predict Cardiovascular and Renal Outcomes in Type 2 Diabetes: A Post Hoc Analysis of the LEADER Trial

Evidence from observational studies and clinical trials in diabetes has demonstrated albuminuria to be a strong predictor of cardiovascular (CV) (1,2) and renal events (3–5). The classification of albuminuria into groups (i.e., normo-, micro-, and macroalbuminuria on the basis of urinary albumin-to-creatinine ratio [UACR] values of 0 to <30 mg/g, 30–300 mg/g, and ≥300 mg/g, respectively) has proven clinically useful to stratify risk and guide treatment decisions. Large meta-analyses have strengthened an emerging body of evidence for the role of albuminuria as a renal risk factor and its reduction as a target for treatment in kidney disease (6,7). In the latter meta-analysis, treatment for the most part was based on inhibition of the renin-angiotensin-aldosterone system (RAAS) or other antihypertensive agents.

Data from a number of trials have indicated that glucagon-like peptide 1 receptor agonists (GLP-1 RAs) lower albuminuria and provide CV and renal benefits in participants with type 2 diabetes (T2D) (8–12). The GLP-1 RAs liraglutide and semaglutide have shown both CV (9) and renal benefits (10) in participants with T2D and high CV risk in the Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) and Semaglutide Unabated Sustainability in Treatment of Type 2 Diabetes (SUSTAIN-6) trials (9,13). In both trials, there was a significant reduction in albuminuria and a prevention of development of macroalbuminuria in the GLP-1 RA–treated groups. In addition, in the Researching Cardiovascular Events With a Weekly Incretin in Diabetes (REWIND) trial, the GLP-1 RA dulaglutide demonstrated an 18% overall reduction in UACR alongside a 15% reduction in the composite renal outcome (new macroalbuminuria, a sustained decline in estimated glomerular filtration rate [eGFR] of ≥30% from baseline, or chronic renal replacement therapy) compared with placebo in a cohort of participants with T2D with and without established CV disease (14). Using the data from the LEADER trial, we tested the hypothesis that a reduction in UACR is associated with a reduction of CV and renal risks in a cohort treated with a GLP-1 RA or placebo on a background of control of established CV risk factors and continuous use of RAAS blockade in the majority of participants.

The LEADER trial design, detailed methods, and primary results have been published previously (9,15). In brief, 32 countries participated in this randomized, double-blind, placebo-controlled trial, which was designed to assess the CV safety of liraglutide in participants with T2D at high CV risk. A total of 9,340 participants were randomized 1:1 to receive either subcutaneous liraglutide (1.8 mg/day or the maximum tolerated dose of 0.6–1.8 mg/day) or matching placebo, both in addition to standard-of-care therapy. The treatment period was 3.5–5 years, with a 30-day follow-up period. The majority (>80%) of the participants were treated with RAAS inhibitors, and >40% received insulin, 88% any glucose-lowering agent, and 76% lipid-lowering agents.

The primary outcome was the time from randomization to first occurrence of a composite of major adverse CV events (MACE) consisting of CV death, nonfatal myocardial infarction, or nonfatal stroke. Secondary time-to-event outcomes included a four-component nephropathy composite (new onset of persistent macroalbuminuria or a persistent doubling of serum creatinine, i.e., confirmed by a second reading [15] and eGFR ≤45 mL/min/1.73 m², need for continuous renal replacement therapy [in the absence of an acute reversible cause], death as a result of renal disease).

In this post hoc analysis, we analyzed the risk of MACE and a three-component nephropathy composite (doubling of serum creatinine and eGFR <45 mL/min/1.73 m², renal replacement therapy, or renal death) in participants with a UACR measurement at baseline and at 1 year after randomization. The component new onset of persistent macroalbuminuria was excluded from the renal composite outcome for this analysis because one subgroup in the current analysis comprised participants with preexisting macroalbuminuria. Participants were stratified into three categories according to change in UACR from baseline to 1 year (>30% reduction, 30‒0% reduction, and any increase from baseline). These thresholds for changes in albuminuria were chosen on the basis of previous analyses of trials using RAAS inhibition (4,16,17). In addition, the analyses were repeated in subgroups with baseline normo-, micro-, and macroalbuminuria. For the purposes of comparison, the group with any increase in UACR from baseline served as the reference.

UACR and serum creatinine levels were measured at randomization, after 12 months and annually thereafter, and at trial completion; additionally, serum creatinine level was measured at month 6. All measurements were done centrally (15). UACR or creatinine measurements less than the limit of quantification (LLoQ) (3.0 mg/L) were imputed using a value of 1/2 × LLoQ; those measurements greater than the higher limit of quantification were imputed using the higher limit of quantification value.

CV and renal events included in the composite outcomes were adjudicated by an independent, blinded committee (15). Time to event from 1 year to end of study according to change in UACR from baseline to the 1-year visit and UACR groups at baseline were analyzed using a Cox proportional hazards model. The analysis was adjusted for treatment allocation alone (liraglutide vs. placebo) as a fixed factor and for treatment and covariates (age, sex, systolic and diastolic blood pressure, eGFR, body weight, glycated hemoglobin [HbA_1c], UACR, and smoking status at baseline and changes in systolic and diastolic blood pressure, eGFR, body weight, and HbA_1c from baseline to the 1-year visit). All participants who underwent randomization and who had measurements of UACR at baseline and at the 1-year visit were included and if there was no event, censored from analysis at time of death or the end of follow-up, whichever came first. Events within the 1st year were excluded from the analysis. Change in UACR at 1 year was analyzed using a mixed-effects model for repeated measures on log-transformed values according to UACR baseline subgroup (normo-, micro-, or macroalbuminuria) adjusted for continuous UACR at baseline (log transformed), age, antidiabetic medication at baseline, sex, and interaction between randomized treatment and UACR subgroup. For each UACR baseline subgroup, the change in continuous UACR from baseline was derived as a ratio (summarized in percentages) according to treatment and across treatment.

We assessed the impact of regression to the mean by calculating the nonparametric regression dilution coefficient using the MacMahon-Peto method, dividing UACR data into deciles (18) (Supplementary Fig. 1). Additionally, we calculated this coefficient using a linear model with the log-transformed UACR values at 1 year as a dependent variable and the log-transformed UACR values at baseline as a covariate and then used the reciprocal of the regression coefficient to estimate the parametric dilution coefficient. These analyses could be potentially affected by a survival bias within the 1st year because patients with a high UACR at baseline risk were at a higher risk for all-cause death, specifically within the 1st year of follow-up and during the trial.

The trial was approved by the participating institutions’ ethics committees/institutional review boards, and all patients provided written informed consent. The trial was conducted in accordance with the Declaration of Helsinki.

Of the 9,340 participants randomly assigned in the LEADER trial, 9,113 had UACR measured at baseline (15). The patient disposition is shown in Supplementary Fig. 2. After 1 year, 8,270 patients (89% of the randomized population) had a follow-up albuminuria measurement and were included in this post hoc analysis. The demographics of this subgroup population are given in Table 1 and did not differ in any notable way from the full study population.

Approximately one-half of the patients had an increase in albuminuria during the 1st year of the trial (n = 4,124; 50% of the population in this analysis) of whom 498 (12.1%) experienced CV events and 113 (2.7%) renal events. The remainder of the population had a reduction of up to 30% from baseline (n = 1,218; 15%) or >30% reduction (n = 2,928; 35%) from baseline UACR during the 1st year. Overall reduction in UACR was 3.5% (95% CI 1, 6), UACR decreased by 15% (13, 18) in the liraglutide group, and there was an estimated increase of 10% (7, 14) in the placebo group. Compared with any increase in UACR (reference), patients with a decrease of up to 30% had a similar risk of MACE (12.1%) with a hazard ratio (HR) of 0.99 (95% CI 0.82, 1.19; P = 0.912). For the composite nephropathy outcome, the HR was 0.97 (0.66, 1.43; P = 0.881). For patients with a 1-year reduction in UACR of >30% from baseline, the HR for MACE was 0.82 (0.71, 0.94; P = 0.006) and 0.67 (0.49, 0.93; P = 0.02) for the composite nephropathy outcome. The associations between early change and subsequent MACE and renal outcomes were consistent in the liraglutide and placebo groups (P for interaction = 0.516 and 0.839 for MACE and renal events, respectively). The data for both end points in the UACR change groups from 1 year and onward tended to favor liraglutide. For the first MACE, the HRs in the UACR change group −30 to 0% and < −30% were 0.89 (0.67, 1.17; P = 0.386) and 0.83 (0.68, 1.01; P = 0.067) in the liraglutide group compared with 1.06 (0.83, 1.36; P = 0.649) and 0.78 (0.64, 0.96; P = 0.021), respectively, in the placebo group. For first renal event, the HRs in the UACR change group −30 to 0% and < −30% were 0.41 (0.22, 0.78; P = 0.007) and 0.46 (0.29, 0.74; P = 0.001) in the liraglutide group compared with 1.01 (0.59, 1.73; P = 0.969) and 0.69 (0.44, 1.10; P = 0.120), respectively, in the placebo group.

In patients with normoalbuminuria at baseline, after 1 year, there was a mean relative reduction in UACR of 14% (95% CI 9, 18). In patients with microalbuminuria, an increase in UACR of 12% (4, 20) was estimated, and in those with macroalbuminuria, UACR more than doubled (121% [92, 153]). Supplementary Fig. 3A shows the unadjusted HRs for MACE by normo-, micro-, and macroalbuminuria subgroups and change in UACR. An albuminuria reduction of >30% from baseline was associated with a reduction in risk of MACE in patients with micro- or macroalbuminuria. The P value for interaction between baseline category and change in UACR adjusted for treatment was 0.26. Figure 1A shows the adjusted HRs for MACE, with a >30% reduction in micro- and macroalbuminuria subgroups significantly associated with lower risk.

Similarly, Supplementary Fig. 3B shows the unadjusted HRs for the composite nephropathy outcome in the same subgroups. Here, a reduction in albuminuria of >30% was associated with renal benefit in patients with macroalbuminuria. The P value for interaction between baseline category and change in UACR adjusted for treatment was 0.89. Figure 1B depicts the same association but with adjusted HRs, and here, a >30% reduction was associated with less renal risk in subgroups with baseline micro- or macroalbuminuria. In addition, a minor reduction from baseline albuminuria was associated with less renal risk. The demographics of the subgroup populations are given in Supplementary Table 1.

The cumulative distribution of change from baseline to 1 year in UACR (logarithm to the ratio between 1-year measurement and baseline) is shown in Supplementary Fig. 4. No interactions were seen between UACR change and use of RAAS inhibitors at baseline according to the three UACR groups for the two end points (data not shown).

Analyses showed that for each SD increase in UACR from baseline (log transformed) to 1 year, the HR was 1.19 (95% CI 1.12, 1.27) for MACE and 1.79 (1.52, 2.12) for renal outcome. A 1% decrease in HbA_1c from baseline to 1 year was associated with change in UACR (β = 0.14; P < 0.001), adjusted for baseline HbA_1c and log-transformed UACR.

Additionally, analyses with relative change in UACR between baseline and 1 year were performed. These adjusted analyses showed that for patients with macroalbuminuria, a doubling of UACR increased the risk of first MACE and risk of renal event by 25% (95% CI 11, 41) and 44% (27, 65), respectively. For patients with normoalbuminuria, the numbers were 0% for both first MACE and first renal event, and for patients with microalbuminuria, the numbers were 1% for first MACE and 3% for first renal event, indicating a very modest risk of UACR increase for these end points. We used clinically relevant changes in UACR used in previous studies (3,19).

Pooled across treatment groups, there was modest evidence of regression to the mean UACR; the nonparametric dilution coefficient (representing regression on the change from baseline) using the MacMahon-Peto method and the parametric dilution coefficient (representing baseline) were 1.24 and 1.22, respectively. Every 1-SD increase in UACR at baseline was associated with a 35% higher risk of first MACE (95% CI 27, 43). Applying the nonparametric dilution coefficient increased this estimate to 43% (34, 56). Correspondingly, every 1-SD increase in baseline UACR was associated with a 3.6-fold (95% CI 4.03, 5.27) higher risk of a renal event, which increased after applying the dilution coefficient to a 5.6-fold (5.63, 7.85) higher risk.

The results of this post hoc analysis from the LEADER trial indicate that a reduction in UACR from baseline to 1 year predicts future benefits on CV and renal outcomes. For example, a >30% reduction of UACR from baseline was associated with a reduced risk of the composite nephropathy outcome. These associations were confirmed after adjusting for clinical variables at baseline and changes in covariates at 1 year. Indeed, approximately one-third of the LEADER population experienced a substantial reduction (i.e., >30%) of UACR, which was seen more frequently with liraglutide than with placebo. Nevertheless, no treatment interaction was observed with the association of change in UACR and MACE or renal outcomes, indicating that the renal benefit of UACR reduction was not restricted to liraglutide-treated patients alone.

In subgroups with micro- or macroalbuminuria at baseline, we found that a 1-year reduction in albuminuria >30% was associated with improved CV and renal outcomes. These findings are reassuring because these subgroups with elevated albuminuria also carry the highest risk of CV and renal events. Any effort to reduce albuminuria should be implemented in routine clinical diabetes care.

These findings from LEADER are of particular interest given that most other evidence of associations of changes in UACR and outcomes have come from trials investigating initiation of RAAS blockade, a well-known mechanism to reduce UACR. In LEADER, the majority of enrolled participants were on standard-dose RAAS blockade at randomization and remained on that therapy for the duration of the trial.

Our findings are in line with previous observational and post hoc studies (19) and meta-analyses (6,7) performed in cohorts where treatment was mostly based on initiation of RAAS inhibitors or non–GLP-1 RA antihypertensive treatments. Of note is the dual benefit associated with a significant reduction of albuminuria for the CV and renal outcomes (4,20). We observed a 25% and 58% relative risk reduction of these outcomes, respectively, in the group with baseline microalbuminuria (on the basis of the adjusted analyses); a >30% reduction in albuminuria after 1 year in the group with baseline macroalbuminuria was associated with a 43% lower risk of both CV and renal outcomes. Few other targeted risk factor interventions in T2D are associated with this magnitude of risk reduction.

Previous post hoc analyses investigating the benefit of albuminuria reduction are mostly from randomized trials of mono- or dual-RAAS–blocking therapies. In an analysis of the Ongoing Telmisartan Alone and in Combination With Ramipril Global End Point Trial (ONTARGET) and Telmisartan Randomized Assessment Study in ACE Intolerant Subjects With Cardiovascular Disease (TRANSCEND) trial, two large CV randomized clinical trials run in parallel in patients with vascular disease or high-risk diabetes, many of whom with albuminuria, Schmieder et al. (20) reported that a twofold or greater decrease in albuminuria predicted both CV and renal benefit compared with a minor change in albuminuria.

Similarly, in the Reduction of End Points in NIDDM With the Angiotensin II Antagonist Losartan (RENAAL) trial, it was demonstrated that a >30% reduction from baseline albuminuria was associated with both greater CV and renal protection (4,21). We chose the same cutoff (>30% reduction) for our analysis, and it is interesting that this is still clinically significant in a modern cohort of patients with T2D, most of whom were on RAAS-blocking treatment. A large meta-analysis of 41 randomized clinical trials demonstrated close associations between albuminuria reduction and lower risk of renal outcome. In the analysis, a 30% decrease in albuminuria was associated with a 27% lower risk for a composite renal end point of end-stage renal disease, eGFR <15 mL/min/1.73 m², or doubling of serum creatinine (6). In addition, in a real-world setting, Italian authors demonstrated that a remission of albuminuria category led to a reduction in renal risk in a cohort gathered from 100 diabetes centers (22). Unfortunately, implementation of albuminuria monitoring remains a challenge, as seen in a cohort study from two U.S. health care systems where only 12% of adults with diabetes and chronic kidney disease had an albuminuria measurement (23).

The novelty of our analysis is that the LEADER trial was not investigating RAAS blockade or antihypertensive treatment but a diabetes treatment with pleiotropic effects. This supports a focus on albuminuria reduction as an overall clinical treatment goal alongside the reduction in glycemic control, blood pressure, and lipid levels in diabetes treatment guidelines (24). The drawback at present is that we are lacking prospective intervention trials that target higher and lower goals of UACR and examine renal and CV outcomes comparable to intensive versus standard goals of blood pressure or glycemic control.

We need a better mechanistic understanding of the potential damage caused by albuminuria in order to develop appropriately targeted therapies. In the meantime, it is comforting that several GLP-1 RAs now have documented albuminuria-lowering effects that may well contribute to their overall renal benefit. Studies have shown reductions in albuminuria of 17–32% with liraglutide (10,11), 2–39% with lixisenatide (25), and 29% with dulaglutide (26). Dipeptidyl peptidase 4 inhibitors, on the other hand, seem to have less albuminuria-lowering potential, as demonstrated in the placebo-controlled Efficacy, Safety & Modification of Albuminuria in Type 2 Diabetes Subjects With Renal Disease With Linagliptin (MARLINA) trial in which linagliptin led to a nonsignificant 6% albuminuria reduction (27). Similar effect sizes were observed in the DELIGHT trial (Albuminuria-Lowering Effect of Dapagliflozin Alone and in Combination With Saxagliptin and Effect of Dapagliflozin and Saxagliptin on Glycaemic Control in Patients With Type 2 Diabetes and Chronic Kidney Disease) where saxagliptin was added to dapagliflozin (28). However, the subsequent linagliptin outcome trial Cardiovascular and Renal Microvascular Outcome Study With Linagliptin in Patients With Type 2 Diabetes Mellitus (CARMELINA) (29) showed potential for reduced albuminuria progression (HR 0.86 [95% CI 0.78, 0.95]; P = 0.003) as did a previous analysis of the Saxagliptin Assessment of Vascular Outcomes Recorded in Patients With Diabetes Mellitus–Thrombolysis in Myocardial Infarction 53 (SAVOR-TIMI 53) trial with saxagliptin, which also showed significant reduction of albuminuria in the normoalbuminuric range (P = 0.021) (30).

There are limitations to our analysis. Although LEADER was a large trial with a long follow-up, this remains a post hoc analysis with all the inherent problems that preclude causal inferences. First, it is not clear whether the reductions in albuminuria are the cause of improved outcomes or merely markers of other factors such as general endothelial integrity. Also, the LEADER trial was conducted in a population with T2D with high CV risk; thus, the findings from this analysis may not be generalizable to a broader patient population. UACR measurement was based on a single urine sample, which may have led to higher variability compared with using two or three samples and potential regression to the mean. However, it has been shown that a single sample can be used in a large study population with T2D and macroalbuminuria (31). Morning spot urine samples are well suited for use in clinical trials of albuminuria, and logistically challenging 24-h urine collections are not needed (32,33). Furthermore, UACR measurements at an earlier stage of treatment, such as after 3 or 6 months of treatment, would have helped to describe the time course of albuminuria changes. Imputation using LLoQ / 2 for urine albumin values could have introduced bias in the normoalbuminuria group such that the decreasing risk of MACE and renal events could have been overestimated, especially in the group with the largest reduction in UACR from baseline to 1 year. This could be due to random drops in UACR caused by this imputation. Finally, no control for multiple comparisons was performed.

In conclusion, the results of the current study in a large, contemporary population of patients with T2D followed for a median of 3.8 years confirm the close association of reductions in UACR with reduced risk for major CV and renal outcomes in patients with T2D at high CV and moderate renal risk. These data strongly support the concept of a randomized controlled trial testing lower and higher target levels of UACR on major CV and renal outcomes.

Acknowledgments. Medical writing and submission support were provided by Aneela Majid and Izabel James of Watermeadow Medical, an Ashfield company, part of UDG Healthcare plc, funded by Novo Nordisk A/S.

Duality of Interest. This analysis was supported by Novo Nordisk A/S. F.P. has received research grants from AstraZeneca, Novo Nordisk, and Novartis; received lecture fees from Novartis, Eli Lilly, MSD, AstraZeneca, Sanofi, Novo Nordisk, and Boehringer Ingelheim; and served as a consultant for AstraZeneca, Bayer, Amgen, Novo Nordisk, and MSD. S.C.B. has received honoraria and teaching and research sponsorship/grants from Abbott, AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Cellnovo, Diartis, Eli Lilly, GlaxoSmithKline, Merck Sharp & Dohme, Novartis, Novo Nordisk, Pfizer, Roche, Sanofi, Schering-Plough, SERVIER, and Takeda and funding fo development of educational programs from Cardiff University, Doctors.net, Elsevier, Onmedica, Omnia-Med, and Medscape. He owns a share of Glycosmedia and has provided expert advice to the All-Wales Medicines Strategy Group and National Institute for Health and Care Excellence U.K. O.M. has served on advisory boards of Novo Nordisk, Eli Lilly, Sanofi, Merck Sharp & Dohme, Boehringer Ingelheim, and AstraZeneca; received grants paid to her institution as study physician by AstraZeneca; received research grant support through Hadassah Hebrew University Hospital from Novo Nordisk; and received speakers’ bureaus honoraria from AstraZeneca, Novo Nordisk, Eli Lilly, Sanofi, Merck Sharp & Dohme, and Boehringer Ingelheim. H.J.L.H. has served as a consultant for AbbVie, Astellas, AstraZeneca, Boehringer Ingelheim, Fresenius, Gilead, Janssen, Merck, Mitsubishi Tanabe, and Mundipharma, with all honoraria paid to his employer. J.F.E.M. has received speaker honoraria from Amgen, AstraZeneca, Boehringer Ingelheim, Braun, Fresenius, Gambro, Eli Lilly, Medice, Novo Nordisk, Relypsa, and Roche; research support from European Union, Canadian Institutes of Health Research, Boehringer Ingelheim, Celgene, Novo Nordisk, Roche, and Sandoz; and consultation fees from AstraZeneca, Bayer, Celgene, Fresenius, Eli Lilly, Lanthio Pharma, Novo Nordisk, Relypsa, Sanifit, and Vifor Pharma. R.P.’s services were paid directly to AdventHealth (formerly Florida Hospital), a nonprofit organization, including consultancy and speaker fees from AstraZeneca, Takeda, and Novo Nordisk; consultancy fees from Boehringer Ingelheim, GlaxoSmithKline, Hanmi Pharmaceutical Co. Ltd., Janssen Scientific Affairs LLC, Ligand Pharmaceuticals, Eli Lilly, Merck, Pfizer, and Eisai; and research grants from Gilead Sciences, Lexicon Pharmaceuticals, Ligand Pharmaceuticals, Eli Lilly, Merck, Sanofi US LLC, and Takeda. I.R. has received personal fees from AstraZeneca, Bristol-Myers Squibb, Boehringer Ingelheim, Concenter BioPharma and Silkim, Eli Lilly, Merck Sharp & Dohme, Novo Nordisk, Orgenesis, Pfizer, Sanofi, SmartZyme Innovation, Panaxia, FuturRx, Insuline Medical, Medial EarlySign, CameraEyes, Exscopia, Dermal Biomics, Johnson & Johnson, Novartis, Teva, GlucoMe, and DarioHealth. T.I., S.R., and B.J.v.S. are Novo Nordisk A/S employees and shareholders. P.R. has received lecture fees from AstraZeneca, Bayer, and Boehringer Ingelheim and served as a consultant for AbbVie, AstraZeneca, Bayer, Eli Lilly, Boehringer Ingelheim, Astellas, Janssen, and Novo Nordisk, all fees given to Steno Diabetes Center Copenhagen, equity interest in Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. F.P. drafted the manuscript, which was revised and approved by all authors, who also assume responsibility for its content. F.P., H.J.L.H., B.J.v.S., and P.R. initiated the study. S.R. performed the analyses, and all authors discussed the analysis plan and results and provided input to the manuscript. All authors had access to the final study results. F.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract and oral form at Kidney Week 2018, San Diego, CA, 23–28 October 2018. Parts of these data were also presented at local congresses by Novo Nordisk affiliates in 2019.