SodiumGlucose Cotransporter-2 Inhibitors and Severe Urinary Tract Infections: Reassuring Real-World Evidence

Abstract

Sodiumglucose cotransporter-2 (SGLT-2) inhibitors, the most recently approved class of antidiabetic drugs, reduce glycemia by inhibiting glucose reabsorption in the renal proximal tubule (1). Through this mechanism of action, these drugs also induce weight loss and confer cardiovascular benefits, including reductions in major adverse events and heart failure (2). As a result, SGLT-2 inhibitors were the first class of antidiabetic drugs to be approved for cardiovascular disease prevention and are now recommended as a preferred second-line therapy among patients with type 2 diabetes and known atherosclerotic cardiovascular disease who cannot attain glycemic control with metformin monotherapy (3). Because of the mechanism of action of SGLT-2 inhibitors, a strong biological rationale indicates that they may increase risk for urinary tract infections (UTIs): The higher glucose concentration in urine with SGLT-2 inhibitor use may promote bacterial growth (4). In 2015, the U.S. Food and Drug Administration (FDA) issued a warning about increased risk for serious UTIs with SGLT-2 inhibitor use (5). This warning was based on 19 cases of urosepsis and pyelonephritis reported to the FDA s Adverse Event Reporting System. Given the underlying UTI risk among patients with type 2 diabetes and the inherent limitations of such systems (for example, no comparator and no denominator), the conclusions that can be drawn from these data have limited strength. Subsequent meta-analyses of data from randomized controlled trials (RCTs) did not find an increased UTI risk with SGLT-2 inhibitors versus placebo overall (relative risk, 1.03 [95% CI, 0.96 to 1.11]). However, molecule-specific analyses suggested that dapagliflozin may be associated with higher risk than placebo (relative risk, 1.23 [CI, 1.03 to 1.46]) (6). Of note, the generalizability of RCT data in this area to everyday clinical practice is unclear. Dave and colleagues (7) examined the association between SGLT-2 inhibitors and risk for severe UTI in a real-world setting. Using 2 large, U.S. databases of insurance claims (MarketScan and Optum), they studied 2 propensity scorematched cohorts to compare the risk for severe UTI associated with SGLT-2 inhibitors versus dipeptidyl peptidase-4 (DPP-4) inhibitors and glucagon-like peptide-1 (GLP-1) receptor agonists. The primary end point was severe UTI, defined as a composite of hospitalizations with sepsis and UTI, those with pyelonephritis, and those with UTI as a primary diagnosis. When data were pooled across databases, SGLT-2 inhibitors were not associated with increased risk for severe UTI compared with DPP-4 inhibitors (hazard ratio [HR], 0.98 [CI, 0.68 to 1.41]) or GLP-1 receptor agonists (HR, 0.72 [CI, 0.53 to 0.99]). Results were similar for the individual components of the primary end point and across several secondary and sensitivity analyses. Molecule-specific analyses showed no important differences between canagliflozin and dapagliflozin compared with DPP-4 inhibitors (HRs, 0.83 [CI, 0.57 to 1.21] and 0.57 [CI, 0.29 to 1.14], respectively) or compared with GLP-1 receptor agonists (HRs, 0.66 [CI, 0.47 to 0.92] and 0.52 [CI, 0.28 to 0.97], respectively). Dave and colleagues study has several strengths. Through its propensity scorematched design, use of active comparators at similar points in the management of type 2 diabetes, and rigorous statistical adjustment (including use of a frailty score), it reduced confounding by indication and by other variables. With more than 100000 patients per cohort, the authors could restrict the analysis to severe UTIs, a clinically important end point among patients with type 2 diabetes. Results were consistent across several analyses, suggesting that the findings are robust to study assumptions. Finally, because the study used real-world data, its findings are more generalizable to everyday clinical practice than those obtained from RCTs in this area. Some limitations are worth noting. As discussed by the authors, the study was restricted to insured persons in the United States, and the generalizability of its findings to uninsured persons and those in other countries is unclear. Furthermore, patients with renal insufficiency, high risk for UTI, or history of UTI were excluded. Although such exclusions may have increased validity, they may also have adversely affected generalizability. They also prevented subgroup analyses among patients at greatest risk for the outcome of interest. With use of an active-comparator, new-user design (8), the study excluded patients who had recently used the comparator drugs; given the dynamic nature of the management of type 2 diabetes, this can result in many excluded patients. Indeed, as shown in Table 1 of Dave and colleagues article (7), about 30% of SGLT-2 inhibitor users had a history of DPP-4 inhibitor use, and the exclusion of such persons can reduce generalizability (a strength of real-world evidence) (9). This represents an ongoing methodological challenge in studies of conditions with escalating, multistep treatments. Finally, despite use of rigorous methods, residual confounding remains possible and probably explains the unexpected protective effects observed in comparisons with GLP-1 receptor agonists. This study has several important implications and represents a key addition to the literature. Although concerns emerged about a potentially increased risk for genitourinary tract infections with use of SGLT-2 inhibitors, accumulating evidence suggests that harms are restricted to genital tract infections (6). Some limitations are noted, but this study is methodologically rigorous and provides reassuring, real-world evidence regarding this potential safety issue. Reassurance, however, comes with some caveats. First, the study excluded high-risk patients and those with a history of UTI, key subgroups for which further evidence is needed. Second, some analyses of secondary outcomes (such as urosepsis) and specific molecules had more modest statistical power, and safety data from RCTs suggest that risk may vary by molecule (6). Safety assessments by large networks, such as the Sentinel System or the Canadian Network for Observational Drug Effect Studies may be needed to conclusively address these issues. Ultimately, although some uncertainty remains, the study by Dave and colleagues (7) provides encouraging evidence of the real-world safety of SGLT-2 inhibitors, allowing patients to benefit from their use with greater confidence in their safety with respect to severe UTI.