The retrospective observational study, reported in Frontiers in Nephrology, emulated the protocol of the phase 2 CONFIDENCE trial. Investigators analysed electronic health records from January 2023 through September 2025 at Deyang Hospital affiliated with Chengdu University of Traditional Chinese Medicine, Shanghai Sixth People’s Hospital affiliated with Shanghai Jiao Tong University School of Medicine, and Southwest Hospital affiliated with Army Medical University. Eligible patients were at least 18 years old and had type 2 diabetes with an estimated glomerular filtration rate of 30 to 90 mL/min/1.73 m², UACR of 100 to less than 5,000 mg/g, serum potassium no higher than 4.8 mmol/L and glycated haemoglobin below 11%. They also had to be receiving stable, maximally tolerated treatment with an angiotensin-converting enzyme inhibitor or angiotensin receptor blocker for at least one month.

Of 9,373 patients identified with both type 2 diabetes and chronic kidney disease, 3,822 met the eligibility criteria and had initiated at least one study drug. A 1:1:1 propensity-score matching procedure then selected 2,895 participants from this eligible group, producing a matched cohort of 965 patients classified under the combination strategy, 965 under finerenone monotherapy and 965 under empagliflozin monotherapy. The assignment analysis of those matched groups was secondary, while the primary causal contrast was estimated through a per-protocol analysis that addressed the bias created by allowing the two drugs to be started sequentially.

The matched cohort had a mean age of 59.6 years, and 62.0% were men. Mean body-mass index was 25.4 kg/m², mean eGFR was 46.4 mL/min/1.73 m² and median UACR was 583 mg/g, with an interquartile range of 313 to 1,124 mg/g. Mean serum potassium was 4.30 mmol/L, mean glycated haemoglobin was 7.7% and mean haemoglobin was 118.6 g/L. Hypertension was present in 93.2%, atherosclerotic cardiovascular disease in 58.8% and diabetic retinopathy in 47.8%. Angiotensin receptor blockers were used by 88.7% of participants and insulin by 75.4%. After matching, 25 of the 27 propensity-score covariates had pairwise standardized mean differences below 0.10, although dipeptidyl peptidase-4 inhibitor use and LDL cholesterol retained residual imbalance, with standardized mean differences of 0.127 and 0.117, respectively.

The primary per-protocol analysis estimated the effect of initiating and sustaining each treatment strategy. Every eligible patient was cloned into all three strategies at time zero. Clones were censored when treatment histories became inconsistent with the assigned strategy, including medication gaps longer than 60 days, switching agents, adding the second study drug to a monotherapy strategy or failing to start the second drug during the combination grace period. Inverse probability of censoring weights, incorporating time-varying potassium, eGFR and systolic blood pressure, were used to adjust for informative censoring. This design was intended to remove the grace-period immortal-time bias inherent in classifying combination therapy from prescriptions that could be filled after time zero.

The per-protocol analysis estimated a day-180 UACR-change difference of −19.5 percentage points for combination therapy versus finerenone, with a 95% confidence interval of −20.5 to −18.5, and −22.8 points versus empagliflozin, with a 95% confidence interval of −23.8 to −21.8. The secondary propensity-score-matched assignment analysis produced differences of −19.6 points versus finerenone, with a 95% confidence interval of −20.5 to −18.7, and −22.9 points versus empagliflozin, with a 95% confidence interval of −23.8 to −22.0. Both comparisons had P values below 0.001 and remained significant after Holm–Bonferroni correction. The assignment estimate was described as an observational analogue of an intention-to-treat comparison, but the authors cautioned that it was not a true intention-to-treat effect because classification of sequential combination treatment used information obtained after time zero.

Secondary albuminuria outcomes were more frequent in the combination group. At day 180, 97.7% of patients receiving combination treatment achieved at least a 30% reduction in UACR, compared with 58.4% receiving finerenone and 45.6% receiving empagliflozin. The respective proportions achieving reductions of at least 40% were 88.3%, 20.5% and 14.3%; those achieving reductions of at least 50% were 55.9%, 3.2% and 1.0%. eGFR initially declined in all three groups, with the largest mean decline in the combination group at day 30, but it partially recovered by day 180. The mean day-180 difference was −1.0 mL/min/1.73 m² versus finerenone, with a 95% confidence interval of −1.3 to −0.7, and −0.4 mL/min/1.73 m² versus empagliflozin, with a 95% confidence interval of −0.7 to −0.1. These early kidney-function estimates require caution because some combination-group patients had not yet received the second drug by day 30.

Adding empagliflozin did not meaningfully reduce the potassium-related risk observed with finerenone. Hyperkalemia, defined as serum potassium above 5.5 mmol/L, occurred in 12.7% of the combination group and 12.8% of the finerenone group, compared with 3.5% in the empagliflozin group. Severe hyperkalemia above 6.0 mmol/L occurred in 1.2%, 1.0% and 0.1% of the respective groups, while hyperkalemia requiring intervention occurred in 4.6%, 4.0% and 1.7%. Acute kidney injury was numerically more frequent with combination treatment, at 4.9%, than with finerenone alone, at 2.8%, or empagliflozin alone, at 2.3%. Symptomatic hypotension occurred in 1.9%, 0.6% and 0.4%, respectively. The study was not powered for formal safety comparisons, and laboratory-defined rates depended on whether follow-up measurements were available.

Serious adverse events occurred in 8.6% of the combination group, 8.1% of the finerenone group and 7.5% of the empagliflozin group. Genital mycotic infection affected 1.6% of the combination group and 1.5% of the empagliflozin group, with none recorded among finerenone-only patients. Diabetic ketoacidosis was uncommon, affecting 0.2%, 0.3% and 0.1%, respectively, while severe hypoglycaemia affected 2.2%, 1.2% and 1.5%. In-hospital mortality ranged from 0.1% to 0.5%, and heart-failure hospitalisation occurred in 1.3%, 1.8% and 2.1%. The investigators noted that patients prescribed finerenone may have received more frequent potassium and creatinine testing, which could have increased the detection of biochemical events in finerenone-containing groups; the magnitude and direction of this surveillance effect could not be quantified.

The findings were consistent across 10 sensitivity analyses and six prespecified subgroups, with no significant interactions. A sensitivity analysis restricted to simultaneous initiation produced per-protocol differences of −20.4 percentage points versus finerenone and −23.8 points versus empagliflozin, close to the primary estimates. A tipping-point analysis also supported the main association when missing day-180 UACR values in the combination group were assigned progressively less favourable values, although the result remained conditional on the assumed pattern of missingness. Loss to follow-up was 29.6% to 32.7% across strategies, and day-180 UACR was missing for 14.7% to 16.5% of participants. Both multiple imputation and censoring weighting relied on unverifiable missing-at-random assumptions, while physician-selected treatment left possible residual confounding.

Overall, the study provides real-world evidence consistent with CONFIDENCE’s short-term albuminuria results in a Chinese tertiary-care population that was younger and had a lower mean eGFR than the original trial population. It does not establish long-term kidney or cardiovascular protection because UACR is a surrogate endpoint and follow-up was limited to 180 days. The findings also cannot automatically be extended to primary care, community settings, unselected Chinese populations or other countries. Their clearest implication is narrower: combination treatment was associated with a substantially larger short-term albuminuria reduction, while hyperkalemia remained frequent in both finerenone-containing groups and acute kidney injury and symptomatic hypotension were numerically more frequent with dual therapy.

Subject of Research: Target trial emulation of finerenone plus empagliflozin for albuminuria reduction in Chinese adults with chronic kidney disease and type 2 diabetes

Article Title: Combination finerenone and empagliflozin for chronic kidney disease with type 2 diabetes: a target trial emulation

Article References: Li, L., Zhang, D., Xu, M., Li, S., Shen, Y., & Lu, S. (2026). Combination finerenone and empagliflozin for chronic kidney disease with type 2 diabetes: a target trial emulation. Frontiers in Nephrology, 6, Article 1930977. https://doi.org/10.3389/fneph.2026.1930977

Image Credits: AI Generated

DOI: 10.3389/fneph.2026.1930977

Keywords: finerenone, empagliflozin, chronic kidney disease, type 2 diabetes, albuminuria, target trial emulation, real-world evidence, hyperkalemia, acute kidney injury, diabetic kidney disease, Combination, chronic