Background: Gout and type 2 diabetes are both common independent risk factors for nephrolithiasis (e.g., 24% of US adults with gout have had nephrolithiasis, vs. 8% without gout[1]). Sodium-glucose cotransporter-2 inhibitors (SGLT2i) increase urinary flow with osmotic diuresis and uric acid excretion (and thereby lower serum urate levels), which could impact nephrolithiasis risk, although potentially in opposing directions. However, nephrolithiasis outcome data for SGLT2i are limited, particularly among the highest risk patients with both gout and type 2 diabetes.

Objectives: To emulate target trials to compare risk of nephrolithiasis among patients with type 2 diabetes (overall and stratified by concomitant gout) initiating SGLT2i versus glucagon-like peptide-1 receptor agonists (GLP1-RA), or dipeptidyl peptidase 4 inhibitors (DPP4i), all contemporary second-line glucose lowering agents.

Methods: We designed and conducted target trial emulations to compare SGLT2i versus GLP1-RA and DPP4i for the risk of kidney stone among adults with type 2 diabetes, including separate target trials comparing these treatments among type 2 diabetes patients with and without gout. We used a Canadian general population database from January 1, 2014 to June 30, 2022, including all dispensed prescriptions. Randomization was emulated using propensity score matching weights based on > 50 pre-exposure prognostic factors including sociodemographics, comorbidities, medications, healthcare utilization, and duration and severity of type 2 diabetes and gout. Kidney stone episodes were ascertained by emergency department (ED), hospitalisation, or outpatient visits (positive predictive value ≥90%[2]). Secondary outcomes included kidney stones resulting in hospitalization/ED visits and those requiring procedures. Weighted Cox proportional hazards models were used to calculate hazard ratios (HR) and rate differences (RD). To evaluate reproducibility and for spurious associations, we also assessed risk of genital infection (positive control outcome, for which we expected SGLT2i would be positively associated), and for risk of any osteoarthritis encounter, a negative control outcome for which we expected a null association.

Results: After propensity score weighting, we included 17,758 initiators of SGLT2i, 17,521 initiators of GLP1-RA and 17,634 initiators of DPP4i (48% male, mean age 59 years). Baseline characteristics were well-balanced between the three treatment groups (standardized differences < 0.1). Rate of kidney stone was lower among those initiating SGLT2i than GLP1-RA or DPP4i (10.7, 16.8, and 14.7 events per 1000 person-years), with HRs of 0.67 (95% CI: 0.59, 0.77) and 0.74 (0.65, 0.83), and rate differences (RDs) of -6.1 (-8.4, -3.7) and -4.0 (-6.1, -1.8) per 1000 person-years, respectively. Among the 8509 gout patients, rates of kidney stone were 16.9, 31.0, and 22.4 events per 1000 person-years, for those initiating SGLT2i, GLP1-RA, and DPP4i, respectively (Figure 1), while those among non-gout patients were 9.7, 15.0, and 13.8 per 1000 person-years. Corresponding HRs among gout patients were 0.60 (0.46, 0.79) and 0.77 (0.60, 0.98), with RDs -14.05 (-22.00, -6.09) and -5.50 (-12.20, 1.19) per 1000 person-years. Compared to gout patients, HRs among non-gout patients were similar (Table 1), whilst RDs were smaller: -5.26 (-7.49, -3.04) and -4.04 (-6.09, -1.99), respectively. Associations persisted regardless of sex, age, or baseline thiazide diuretic use (Table 1). The protective associations also persisted for kidney stones requiring ED visits/hospitalizations or procedures, whereas SGLT2i initiators showed higher risk of genital infection, and no altered risk of osteoarthritis encounter, compared with GLP1-RA or DPP4i, both as expected (Table 1).

Conclusion: The nephrolithiasis benefits associated with SGLT2i in these target trial emulations may guide selection of glucose-lowering therapy in gout patients with type 2 diabetes, among whom the absolute risk benefits associated with SGLT2i use are large.

REFERENCES: [1] Zhu et al. Am J Med 2012.

[2] Semins et al. J Urol 2010.

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

Acknowledgements: NIL.

Disclosure of Interests: Natalie McCormick: None declared, Chio Yokose: None declared, Leo Lu: None declared, Deborah Wexler Data Monitoring Committees for Novo Nordisk, J. Antonio Aviña-Zubieta: None declared, Mary De Vera: None declared, Chixiang Chen: None declared, Rozalina McCoy Emmi, personal fees for the development of patient education materials about diabetes; Yale New Haven Health System, personal fees for the development of quality measures related to diabetes, Gary Curhan Chief Medical Officer at OM1, Inc, GSK, Hyon Choi Ani, LG, Horizon, Shanton, and Protalix, Horizon.