Reducing Kidney Disease Burden in Type 2 Diabetes with SGLT2 Inhibitors

The past decade has seen a rise in available therapies to treat type 2 diabetes mellitus (T2D) that have demonstrated significant reductions in adverse cardiovascular and kidney outcomes, independent of glycemic control.¹ Among these agents, the sodium-glucose cotransporter inhibitors (SGLT2i) have emerged as a bedrock of cardio-kidney therapies, demonstrating consistent benefits across a wide range of baseline Kidney Disease Improving Global Outcomes (KDIGO) risk categories for CKD. Randomized controlled trials (RCTs) such as CREDENCE, DAPA-CKD, and EMPA-KIDNEY have established and replicated the kidney and cardiovascular outcomes benefits with SGLT2i in CKD and provide the evidence base for SGLT2i as a foundational class of therapies in a majority of patients with diabetic and nondiabetic CKD.² In addition, post hoc analyses from cardiovascular outcomes trials such as EMPAREG OUTCOME,³ DECLARE TIMI 58,⁴ and the CANVAS program⁵ have demonstrated reductions in albuminuria progression, which is important for upstream targeting of the natural trajectory of CKD in patients with T2D.

Despite the robust body of data from RCTs of SGLT2i supporting their cardio-kidney benefits, questions remain about their utilization in the care pathway of a patient with T2D upfront, with primary prevention of kidney disease as the intent. Relying on RCTs for guidance in this area is limited by the need to enroll larger numbers of a relatively lower-risk group of individuals with T2D, with longer durations of follow-up required to accrue sufficient clinical outcomes of interest. In this context, thoughtfully performed observational studies drawing from real-world databases have the ability to shed light on the effectiveness (and safety) of SGLT2i outside of a trial setting and quantify potential benefits in lower-risk patient populations who would not have qualified for inclusion in the current pool of SGLT2i RCTs.

In this issue of CJASN, Melzer Cohen et al.⁶ describe a real-world comparison of SGLT2i and dipeptidyl peptidase-4 inhibitors (DPP4i) on kidney outcomes in patients with T2D and in the subgroup without established cardiovascular or kidney disease, in particular. The advantages of this cohort from a single health care system within Israel include a large size dataset with comprehensive health variable acquisition, longer follow-up duration compared with other real-world registry reports,⁷ and very low (≤1%) data attrition rates. A propensity score-matched analysis was used to balance baseline characteristics across 90 variables. The primary outcomes included a composite kidney-specific outcome (≥40% reduction from baseline eGFR or new kidney failure) and a kidney-or-death outcome (kidney-specific outcome or all-cause mortality). Categorical changes in the urine albumin-to-creatinine ratio (UACR) for patients with values <300 mg/g at inclusion and eGFR slopes were also analyzed. The subgroup of patients with baseline low cardio-kidney risk was defined by the lack of a history of cardiovascular disease, albuminuria, or an eGFR <60 ml/min per 1.73 m². Both intention-to-treat (ITT) and as-treated analyses were performed.

After balancing, the comparator cohort contained 9824 individuals who were started on an SGLT2i (empagliflozin or dapagliflozin), and 9824 individuals were started on a DPP4i (sitagliptin, linagliptin, vildagliptin, and saxagliptin). Over half of each group had no evidence of cardiovascular or kidney disease. Baseline eGFR was 89 and 90 ml/min per 1.73 m² in the SGLT2i and DPP4i groups, respectively. Median follow-up time for the entire cohort was 38 months, which is substantially longer than previously reported real-world analyses.⁷ For the ITT, the kidney-specific outcome rate was 6.9 versus 9.5 events per 1000 patient-years for SGLT2i versus DPP4i (hazard ratio [HR], 0.72; 95% confidence interval [CI], 0.61 to 0.86). The kidney-or-death outcome demonstrated similar benefits with SGLT2i versus DPP4i (HR, 0.80; 95% CI, 0.72 to 0.89). In the subgroup of patients without evidence of cardiovascular or kidney disease, initiation of SGLT2i was associated with a lower risk for the composite kidney-or-death outcome (HR, 0.77; 95% CI, 0.61 to 0.97), but not for the kidney-specific outcome (HR, 0.67; 95% CI, 0.44 to 1.02) The as-treated analysis was overall consistent with the ITT analysis.

The rate of eGFR decline in the entire population was −1.52 versus −2.01 ml/min per 1.73 m² per year for SGLT2i versus DPP4, with a between-group difference of 0.49 ml/min per 1.73 m² (0.35–0.62), and 0.48 ml/min per 1.73 m² (0.32–0.64) in the subgroup with no cardiovascular or kidney disease. The mitigated benefit seen in the highest albuminuria and lowest eGFR categories is likely related to the low number of patients in these subgroups. An important advantage of this analysis as compared with previous real-world comparisons⁷ is the availability of UACRs and the evaluation of categorical changes in albuminuria as an outcome. In the current analysis, worsening of albuminuria was marginally lower in the ITT and significantly lower in the as-treated groups for SGLT2i versus DPP4i, with the caveat that the DPP4i have shown reductions in albuminuria in trials such as CARMELINA, but not in others, such as MARLINA-T2D.⁸

Important nuances in this analysis deserve attention. First, a noted limitation is the use of propensity score-matching in observational studies. Despite the growing use of propensity scores to mimic RCT characteristics for observational studies, this technique introduces restrictions within the populations analyzed that limit generalizability of the findings. In this analysis, the population was reduced by 50% after propensity score-matching. Imputation of missing values using the corresponding mean values for each group may have introduced bias that associates with the outcomes. Second, there was nearly two-thirds discontinuation or drug class switch in this cohort. To the investigators' credit, a full accounting of the outcomes associated with this important censoring category was provided in Supplemental Table 3. When examining the signal for the kidney-or-death outcome, it is important to note that while DPP4i were chosen by the investigators for their neutral effects on eGFR, they do not have neutral cardiovascular effects. Saxagliptin and alogliptin have Food and Drug Administration warnings for serious heart failure events, and DPP4i, as a class, have shown a fairly consistent signal for worsening heart failure.⁹ Thus, the mortality benefits shown in this analysis have to be interpreted with the context that DPP4i are not neutral comparators for cardiovascular adverse event risk. Third, this analysis represents a very specific population; thus, the findings from this analysis should be externally validated by analyses with more diverse racial and ethnic characteristics. Fourth, safety data are not reported in this analysis. Safety data are particularly important when considering upstream and prolonged use of SGLT2i for kidney disease prevention, given that all the SGLT2i kidney outcomes trials were stopped early for efficacy, with limited insights into long-term safety signals. Finally, an effect modification was seen for baseline renin-angiotensin system inhibitor (RASi) use, with greater benefit with SGLT2i when RASi were not used. This observation is helpful in informing care for patients with T2D who are not on RASi for any underlying reason, without taking away from the foundational role of RASi in albuminuric CKD.

The utilization of eGFR chronic slope in this analysis deserves mention. Glomerular filtration rate slope is an emerging candidate as a surrogate end point for CKD trials. The use of chronic eGFR slope changes in this analysis intends to minimize the effects from the expected, early, and hemodynamically mediated acute eGFR decline with SGLT2i use and capture a potentially more meaningful signal over the longer follow-up duration available. However, there are limitations to this approach even when used in an RCT setting, wherein using postrandomization values as baseline will result in loss of the protection afforded by randomization. The magnitude of the effect of therapies on eGFR slope after excluding the initial acute decline is less certain especially in slow progressors, which affects the ability to predict hard clinical kidney end points. Nonetheless, when examining the potential for therapies such as SGLT2i in low-risk populations and certainly in the domain of primary prevention for CKD, there is a definite place for using eGFR slope as an outcome, especially with emerging data on the utility of total eGFR slope.¹⁰ It is noteworthy that a comparable analysis from CVD-REAL 3 matched patients by eGFR decline rates before the initiation of SGLT2i or other glucose-lowering therapies, which is a stronger method to ensure more equitable distribution of baseline CKD risk.⁷

The findings from this analysis add to a growing body of real-world data that confirm the efficacy of SGLT2i across a broad spectrum of cardio-kidney risk and raise hope for defining clinical contexts and phenotypes of patients with T2D where early initiation of these agents will provide a favorable risk-benefit profile. There is an unmet need for global cardio-kidney-metabolic risk assessment beyond CKD prevention, that factors in social determinants of health, to ensure that patients with T2D who will benefit the most from these valuable therapies interface with them in an equitable fashion. This will allow for major inroads to be made to ensure heart, kidney, and metabolic health for all patients with diabetes, on a global scale.

Disclosures

R.O. Mathew reports employment with Loma Linda VA HCS; consultancy for Data Safety Monitoring Committee for device company–sponsored clinical trial—Procyrion; and ownership interest in Apple, Corning, Lowes, and Snowflake. J. Rangaswami reports employment with Washington DC Veterans Administration Medical Center; consultancy for Bayer, Boehringer-Lily, and Edwards LifeSciences; advisory or leadership role for Medical Advisory Board for Procyrion Inc (Aortix); speakers bureau for Boehringer-Ingelheim/Lily; and other interests or relationships as Chair, Council on the Kidney in Cardiovascular Disease, AHA.

Funding

None.

Author Contributions

Conceptualization: Janani Rangaswami.

Writing – original draft: Roy O. Mathew, Janani Rangaswami.

Writing – review & editing: Roy O. Mathew, Janani Rangaswami.