Patients with type 2 diabetes who initiated sodium-glucose cotransporter 2 inhibitors had a lower incidence of kidney cancer than matched patients who initiated dipeptidyl peptidase-4 inhibitors, according to a population-based cohort study.
The retrospective findings should be interpreted as hypothesis-generating and do not establish that sodium-glucose cotransporter 2 inhibitors prevent kidney cancer. The researchers reported that the association "needs to be further validated," and the absolute number of kidney cancer events was small.
Researchers used the Korean National Health Insurance Service database to identify patients aged 20 to 79 years who started oral glucose-lowering drugs for type 2 diabetes from 2014 to 2018. Eligible patients had no prior kidney cancer diagnosis, had complete information, and had available health examination data. Following 1:1 propensity score matching, the primary analysis included 57,103 new sodium-glucose cotransporter 2 (SGLT2) inhibitor users and 57,103 new dipeptidyl peptidase-4 (DPP-4) inhibitor users.
The active-comparator, new-user design was intended to compare patients with similar treatment-stage diabetes. The researchers noted that DPP-4 inhibitors are commonly used as second-line or later oral glucose-lowering therapy in Korea and have been reported to be neutral for kidney cancer risk.
SGLT2 inhibitor use was defined as initiation of dapagliflozin, empagliflozin, ertugliflozin, or ipragliflozin with at least 90 cumulative days of supply during follow-up. DPP-4 inhibitor use was defined similarly for drugs in that class.
The primary outcome was incident kidney cancer, defined as a first hospitalization with an ICD-10 kidney cancer code and confirmation by a cancer registration code for out-of-pocket payment deduction.
During 408,481 person-years of follow-up, 102 kidney cancer events occurred. Mean follow-up was 3.5 years among SGLT2 inhibitor users and 3.6 years among DPP-4 inhibitor users. Kidney cancer occurred in 38 SGLT2 inhibitor users and 64 DPP-4 inhibitor users, corresponding to incidence rates of 18.9 and 30.8 cases per 100,000 person-years, respectively. The absolute rate difference was approximately 12 fewer cases per 100,000 person-years of follow-up.
SGLT2 inhibitor use was associated with a 40% lower hazard of kidney cancer compared with DPP-4 inhibitor use. The hazard ratio was 0.60, with a 95% confidence interval of 0.40 to 0.89. The association was similar following additional adjustment for the covariates used in propensity score matching.
The association also remained generally consistent across subgroups stratified by age, sex, body mass index, smoking status, hypertension, lipid-lowering drug use, insulin use, metformin use, sulfonylurea use, and diabetes-related complications, with no statistically significant interaction reported.
Several sensitivity analyses were generally consistent with the primary finding. Results were not materially changed when the minimum 90-day supply requirement was removed, an analysis intended to address potential immortal time bias. The lower kidney cancer risk estimate also persisted with multivariable adjustment without propensity score matching, inverse probability of treatment weighting, and 1:3 propensity score matching.
In a secondary analysis comparing SGLT2 inhibitor users with matched nonusers of SGLT2 inhibitors, use of the drug class was associated with a 34% lower hazard of kidney cancer. The hazard ratio was 0.66, with a 95% confidence interval of 0.49 to 0.89. In additional analyses, SGLT2 inhibitor use was not associated with a higher risk of bladder cancer or breast cancer compared with DPP-4 inhibitor use.
The researchers proposed several possible mechanisms for the association, including effects on glucose uptake in renal tumor cells, ketogenesis, autophagy, and glomerular hyperfiltration. However, these mechanisms were not directly tested in the cohort study. The absence of tumor pathology data also prevented the researchers from determining whether the association was concentrated in renal cell carcinoma or in specific subtypes such as clear cell renal cell carcinoma, where some of the proposed mechanisms may be more biologically plausible.
The study had several limitations. The observational design leaves open the possibility of residual confounding, including confounding related to prescribing patterns, glycemic control, and other unmeasured clinical factors. A1C and postprandial glucose data were not available, and changes in key clinical variables over time could not be captured.
The relatively short follow-up period is another important limitation for a cancer outcome. Some kidney cancers diagnosed during follow-up may have been present but undetected at baseline, particularly given the slow-growing pattern of some renal tumors. Differences in baseline health status or health care utilization could also have influenced incidental cancer detection, although the direction of any such bias is uncertain.
The study also lacked individual SGLT2 inhibitor agent and dose data, preventing assessment of dose response or whether the association differed across the agents included in the analysis.
Generalizability may be limited because the cohort included only Korean patients and reflected Korean prescribing and reimbursement patterns. The researchers noted that glucagon-like peptide-1 receptor agonist use accounted for less than 0.1% of the study population, partly because of reimbursement restrictions during the study period. That prescribing context may differ substantially from current U.S. practice, where glucagon-like peptide-1 receptor agonists are more widely used and may influence cardiometabolic and kidney risk profiles.
Disclosures can be found in the published study.
Source: BMC Medicine
