Empagliflozin versus DPP4i or GLP ‐ 1RA for the risk of nephrolithiasis in patients with type 2 diabetes: Research letter from the EMPRISE cohort study

Adults with diabetes and chronic hyperglycemia are at an elevated risk of developing kidney stones.1-3 While a meta-analysis of randomized clinical trials suggested no association between sodium-glucose cotransporter-2 inhibitors (SGLT2i) and nephrolithiasis,4 post hoc analysis of the EMPA-REG OUTCOME trial and recently published observational studies showed a decreased risk of nephrolithiasis events among SGLT2i users compared with users of glucagon-like peptide-1 receptor agonists (GLP-1RA) or dipeptidyl peptidase 4 inhibitors (DPP4i).5-7 However, to date, no studies have evaluated the association between empagliflozin, the most commonly prescribed SGLT2i,8 and the risk of kidney stones utilizing clinical practice data from a broad US population. Therefore, as part of the EMPagliflozin comparative effectIveness and SafEty (EMPRISE) study,9 we evaluated the association between the initiation of empagliflozin versus DPP4i or GLP-1RA on the risk of nephrolithiasis in patients with type 2 diabetes (T2D). For this longitudinal cohort study, we utilized deidentified data from two US-based commercial, administrative health care claims databases (Optum's de-identified Clinformatics® Data Mart Database and IBM® Marketscan®) and Medicare fee-for-service (FFS). These datasets contain information on demographics, inpatient and outpatient diagnosis, procedures, and outpatient pharmacy dispensing records. Mass General Brigham's Institutional Review Board approved this study, and data licence agreements were in place for all databases. In accordance with the EMPRISE study design,9 we generated two separate new-user, active-comparator, sequentially built cohorts (empagliflozin vs. DPP4i and empagliflozin vs. GLP-1RA). In each cohort, the study population included adults with T2D aged ≥18 years (≥66 years in Medicare FFS) who initiated study drugs between 1 August 2014 and 30 September 2019, and had at least 365 days of continuous enrollment in the databases prior to the cohort entry date, that is, the date of the first pharmacy dispensing after a washout period of 365 days. Patients with a history of nephrolithiasis diagnosis at any time before cohort entry, a diagnosis of end-stage kidney disease, solid organ transplant, human immunodeficiency virus, type 1 diabetes, secondary or gestational diabetes, or nursing home admission during the baseline period were excluded (Figure S1). Participant follow-up started a day after cohort entry and continued until the earliest occurrence of nephrolithiasis, death, medication switch (including switching within a class) or discontinuation, health plan disenrollment, or end of study period (30 September 2019). The outcome of interest was nephrolithiasis identified using claims codes in the inpatient or outpatient setting (Figure 1).10 We conducted a secondary outcome analysis using a more stringent outcome definition by requiring a diagnosis as well as a nephrolithiasis-related procedure code within 90 days of that initial diagnosis. Covariates were selected a priori using expert knowledge and published literature and were measured during the 365-day baseline period and on the cohort entry date. They included demographics, summary measures of frailty and disease burden, proxies of diabetes severity, antidiabetic and non-antidiabetic medication use, comorbidities, and measures of healthcare utilisation. To adjust for confounding and address channelling over the study period, we constructed propensity score models for each cohort using all measured covariates within strata of baseline cardiovascular disease status and time block (i.e., time 1: August 2014–September 2017 and time 2: October 2017– September 2019), separately in each database. Patients were 1:1 propensity score matched (PSM) within each stratum and database using the nearest neighbour methodology with a maximum caliper width of 0.01 of the PS. For each cohort, we calculated the number of nephrolithiasis event rates per 1000 person-years with a 95% confidence interval (CI), stratified by exposure group. We estimated hazard ratios (HRs) with 95% CIs using Cox proportional hazards regression models and absolute rate differences (RD) with 95% CIs in pre- and post-matched cohorts. We pooled database-specific results using a fixed-effects meta-analysis to obtain effect estimates in the overall population. We plotted Kaplan–Meier curves to display the cumulative incidence of nephrolithiasis over the follow-up period. We conducted subgroup analyses by age (<65 years, ≥65 years), presence of heart failure and body mass index (BMI) (<30 kg/m2, ≥30 kg/m2). Overall, after PS matching, we identified 102 148 matched pairs for the empagliflozin versus DPP4i cohort and 130 125 matched pairs for the empagliflozin versus GLP-1RA cohort. Across cohorts, each covariate was balanced between exposure groups, as demonstrated by standardized differences of <10%. In both cohorts, the mean age was 62 years, ~46% were female, 71% were White and 39% had obesity-related codes. Selected PSM baseline characteristics for both cohorts are presented in Table 1. Over a mean follow-up period of ~8 months on treatment, in cohort 1, there were 1087 nephrolithiasis events in the empagliflozin group and 1428 events in the DPP4i group, corresponding to HR of 0.75 (95% CI 0.69, 0.81) and RD of −5.42 (95% CI −6.84, −4.00)/1000 PY. For cohort 2, there were 1399 events among empagliflozin initiators and 1825 events among GLP-1RA initiators, corresponding to HR of 0.71 (95% CI 0.66, 0.76) and RD of −6.48 (95% CI −7.75, −5.21)/1000 PY. In the secondary analysis using a more stringent outcome definition, the direction of the effect was consistent (Figure 1). Findings from subgroup analyses by age (<65, ≥65 years) and BMI categories were consistent; however, within a subgroup of patients with heart failure, findings were attenuated towards the null in both cohorts (Figure 1). Cumulative incidence curves for the primary analysis diverged after 3 months of treatment initiation (Figure S2). In this large, nationwide cohort study of adults with T2D, initiation of empagliflozin was associated with a decreased risk of nephrolithiasis when compared with initiators of either DPP4i or GLP-1RA. Findings from predefined subgroups and secondary analyses were mostly consistent with results from the primary analysis. Previous studies evaluated the effect of the entire SGLT2i class,5, 6 to our knowledge, this is the first study investigating the association between empagliflozin initiation and nephrolithiasis in patients with T2D utilizing data collected from routine clinical practice. Currently, the available evidence on the association between empagliflozin and nephrolithiasis was mainly generated from a post hoc pooled analysis of 20 phase I-IV randomized placebo-controlled trials, including the EMPA-REG OUTCOME trial.7 In this retrospective analysis, there were 104 urolithiasis events among empagliflozin users compared with 79 in the placebo group, corresponding to an approximately 40% reduced risk. However, events were captured based on adverse event reports of site investigators during trial follow-up rather than adjudication of cases through chart review. Moreover, nephrolithiasis was not a pre-specified outcome. In terms of studies that evaluated the overall drug class, two published observational studies found that SGLT2i use was associated with a reduced risk of nephrolithiasis.5, 6 Most recently, we evaluated the association between the entire SGLT2i class and nephrolithiasis and found a 26% reduction in risk compared with DPP4i (HR: 0.74, 95% CI [0.71, 0.77]) and a 31% reduction in risk when compared with GLP-1RA (HR: 0.69, 95% CI [0.67, 0.72]). The potential biological mechanism for the observed risk reduction associated with SGLT2i use could be due to increased urinary citrate excretion and reduced stone formation.6 However, the potential beneficial effect of empagliflozin, the most selective SGLT2i marketed in the United States,11 has not yet been evaluated. Our study confirmed results from previous publications. Results from our subgroup analyses were mostly consistent with findings from the primary analysis, except in patients with heart failure, in which our study suggests an attenuated effect of empagliflozin on nephrolithiasis in both cohorts. Our study has several strengths, including its large cohort size and nationwide scope increasing the generalizability of our findings. We also used two active comparator classes that are prescribed at similar T2D stage and adjusted for a large number of measured confounders and proxies, hence reducing the potential for confounding by indication. Nevertheless, we cannot entirely rule out residual confounding. Another limitation of our study is the relatively short mean follow-up on treatment (~8 months), which is consistent with real-world chronic medication adherence patterns.12, 13 Despite excluding patients with a recorded history of kidney stones, we cannot entirely ascertain whether all the cases were incident and not recurrent since we used diagnosis codes to identify events. However, we expect any outcome misclassification to be non-differential between exposure groups. Finally, it should be noted that some of the subgroups, such as patients with heart failure, included a relatively small number of patients, and as a result, they may be underpowered to quantify potential differences between exposure groups with precision. In conclusion, in this large cohort study, the initiation of empagliflozin, the most selective and most frequently prescribed SGLT2i, was associated with a decreased risk of nephrolithiasis compared with the initiation of DPP4i or GLP-1RA. Our study helps inform clinical decision-making in patients with T2D and at an elevated risk of developing nephrolithiasis. HT, JMP and EP were involved in the study conception and design. HT and EP conducted the statistical analyses. All authors were involved in the interpretation of results. HT wrote the first draft of the manuscript, and all authors edited, reviewed and approved the final version of the manuscript. EP obtained funding for the study. EP supervised the study, and HT and EP are the guarantors of this work. All authors approved the final version of the manuscript and agreed to be accountable for the accuracy of the work. None. This study was supported by a research grant to the Brigham and Women's Hospital from Boehringer Ingelheim. The funders had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript; and decision to submit the manuscript for publication. The authors had full control of the design and conduct of the study and interpretation of the study's findings. HT, JMP and HZ have no conflicts of interest to disclose. EP was supported by research grants from the Patient-Centered Outcomes Research Institute (DB-2020C2-20 326), the Food and Drug Administration (5U01FD007213), and the National Institute of Diabetes and Digestive and Kidney Diseases (R01DK138036), not related to the topic of this work. DJW reports serving on Data Monitoring Committees for Novo Nordisk. NS is an employee of Boehringer-Ingelheim. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.16303. A data use agreement is required for each of the data sources. Our data use agreements do not permit us to share patient-level source data or data derivatives with individuals and institutions not covered under the data use agreements. The databases used in this study are accessible to other researchers by contacting the data providers and acquiring data use agreements/licenses. Data S1. Supporting information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.