Empagliflozin lowers serum uric acid in chronic kidney disease: exploratory analyses from the EMPA-KIDNEY trial

Abstract

Background: Hyperuricaemia and gout are common in chronic kidney disease (CKD). We aimed to assess the effects of sodium-glucose co-transporter-2 (SGLT2) inhibition on uric acid (urate) and gout in patients with CKD.

Methods: The EMPA-KIDNEY trial randomised 6609 patients with CKD to receive either empagliflozin 10 mg daily or matching placebo over a median of 2 years of follow-up. Serum uric acid was measured at randomisation then at 2 and 18 months of follow-up and the effects of empagliflozin were analysed using a pre-specified mixed model repeated measures approach. Participant-reported gout events were analysed in Cox regression models (first events) with the Andersen-Gill extension (total events). A post hoc composite outcome included new initiation of uric acid-lowering therapy or colchicine. EMPA-KIDNEY primary and kidney disease progression outcomes were also assessed in subgroups of baseline serum uric acid.

Results: Baseline mean ± standard deviation serum uric acid concentration was 431 ± 114 µmol/l. Allocation to empagliflozin resulted in a study-average between-group difference in serum uric acid of -25.6 µmol/l [95% confidence interval (CI) -30.3 to -21.0], with larger effects in those with higher eGFR (trend P < .001) and without diabetes (heterogeneity P < .001). Compared with placebo, empagliflozin did not significantly reduce first or total gout events [hazard ratio 0.87 (95% CI 0.74-1.02) for the 595 first events and 0.86 (0.72-1.03) for the 869 total events] with similar hazard ratios for the post hoc composite and across subgroups, including by diabetes and eGFR. The effect of empagliflozin on the primary outcome and kidney disease progression outcomes were similar irrespective of the baseline level of uric acid.

Conclusions: SGLT2 inhibition reduces serum uric acid in patients with CKD, with larger effects at higher eGFR and in the absence of diabetes. However, the effect on uric acid is modest and did not translate into reduced risk of gout in EMPA-KIDNEY.

Keywords: CKD; SGLT2 inhibitor; empagliflozin; gout; uric acid.

Graphical Abstract

Figure 1:

Effect of empagliflozin versus placebo on serum uric acid overall. *To convert uric acid to mg/dl, divide by 59.48 (25.6 µmol/l ≈ 0.4 mg/dl). Analysis required participants to have at least one measurement of uric acid during follow-up at 2 and/or 18 months (n = 2691 participants). The value at time 0 is the overall mean of baseline values in all analysed participants in the empagliflozin and placebo arms combined. Study-average differences between treatment groups (empagliflozin minus placebo) are derived from an MMRM adjusted for baseline serum uric acid (in continuous form), baseline × time interaction, the covariates used in the minimisation algorithm (categories of age, sex, diabetes, eGFR, uACR and region), fixed categorical effects of time, treatment allocation and treatment × time interaction and weighted in proportion to the amount of time between follow-up visits. Missing baseline uric acid (4/2691 participants) was imputed with the baseline mean.

Figure 2:

Effects of empagliflozin versus placebo on serum uric acid in subgroups defined by baseline characteristics. *To convert uric acid to mg/dl, divide by 59.48 (380 µmol/l ≈ 6.4 mg/dl; 480 µmol/l ≈ 8.1 mg/dl). Analysis required participants to have at least one measurement of uric acid during follow-up at 2 and/or 18 months (n = 2691 participants). Absolute differences in study-average uric acid between treatment groups (empagliflozin minus placebo) are derived from an MMRM adjusted for baseline serum uric acid (in continuous form), baseline × time interaction, the covariates used in the minimisation algorithm (categories of age, sex, diabetes, eGFR, uACR and region), fixed categorical effects of time, treatment allocation and treatment × time interaction and weighted in proportion to the amount of time between follow-up visits. Interaction terms are included in the MMRM to assess for heterogeneity (sex, diabetes, previous gout, any diuretic therapy) between or trend (eGFR, serum uric acid) across subgroup-specific means and standard errors. Relative differences are presented in Supplementary Fig. 4. Uric acid–lowering therapy includes xanthine oxidase inhibitors and primary uricosuric drugs (see Methods). Missing baseline uric acid (4/2691 participants) was imputed with the baseline mean and thus participants are included in the middle subgroup category (≥380–<480 µmol/l).

Figure 3:

Effects of empagliflozin versus placebo on first and recurrent gout events in subgroups. Total (first and recurrent) self-reported episodes of gout recorded as adverse events (serious or non-serious) analysed using the Andersen–Gill extension of Cox regression with adjustment for the covariates used in the minimisation algorithm (categories of age, sex, diabetes, eGFR, uACR and region). n/N represents the number of events/the number of participants and rate is expressed as the total events per 1000 patient-years. Reasons for missing uric acid: analyses were not conducted in participants from China (n = 986) or participants who did not provide additional optional consent for long-term sample storage for biochemical analysis. *To convert uric acid to mg/dl, divide by 59.48 (380 µmol/l ≈ 6.4 mg/dl; 470 µmol/l ≈ 7.9 mg/dl). ^†Trend test excludes missing category. Uric acid–lowering therapy includes xanthine oxidase inhibitors and primary uricosuric drugs (see Methods).

Figure 4:

Effects of empagliflozin versus placebo on the primary outcome and kidney disease progression by serum uric acid concentration at randomisation. *To convert uric acid to mg/dl, divide by 59.48 (380 µmol/l ≈ 6.4 mg/dl; 470 µmol/l ≈ 7.9 mg/dl). Progression of kidney disease is defined as ESKD (initiation of maintenance dialysis or kidney transplant), a sustained decrease in eGFR to <10 ml/min/1.73 m², a sustained decrease in eGFR of at least 40% from baseline or death from renal causes. Reasons for missing uric acid: analyses were not conducted in participants from China (n = 986) or participants who did not provide additional optional consent for long-term sample storage for biochemical analysis. P_trend tests for trend across the three non-missing categories.