Abstract
BACKGROUND: Given the cardiovascular, renal, and survival benefits of GLP-1 receptor agonists for diabetes, these agents could be effective among kidney transplant recipients. However, kidney transplant recipients are distinct from GLP-1 receptor agonist trial participants, with longer diabetes duration and severity, greater end-organ damage, increased cardiovascular risk, and multimorbidity. We examined GLP-1 receptor agonist real-world effectiveness and safety in kidney transplant recipients with diabetes. METHODS: This USA-based retrospective cohort study included kidney transplant recipients with type 2 diabetes at transplantation and Medicare as their primary insurance from a national registry linked with Medicare claims. Post-transplantation GLP-1 receptor agonist use was identified through Medicare claims. Death-censored graft loss was estimated using the Fine-Gray sub-distribution hazard model and extended Cox models were used for mortality and safety endpoints. Models incorporated inverse probability of treatment weights. To further test whether bias could affect the main results, a cohort was created in which each GLP-1 receptor agonist user was matched with a kidney transplant recipient who had not started a GLP-1 receptor agonist, was alive with a functioning graft, and had accrued the same amount of post-transplant survival time. FINDINGS: Between Jan 1, 2013 and Dec 31, 2020, we identified 44 536 first time kidney transplant recipients with Medicare as primary payer in the 6 months before and at transplantation. 24 192 patients were excluded as they did not have type 2 diabetes. 2328 patients were ineligible (1916 had missing values and 412 used GLP-1 receptor agonists before transplantation). The primary cohort thus consisted of 18 016 kidney transplant recipients with diabetes. Of these patients, 1969 (10·9%) had at least one GLP-1 receptor agonist prescription filled post-transplant. Compared with patients who had not received a GLP-1 receptor agonist, GLP-1 receptor agonist users were younger (median age at transplant 57 years [IQR 49-64] vs 60 years [51-66], p<0·0001) and more likely to be female (786 [39·9%] vs 5645 [35·2%], p<0·0001). Among GLP-1 receptor agonist users, 552 [28·0%] were non-Hispanic White, 703 [35·7%] were non-Hispanic Black, and 568 [28·8%] were Hispanic. The 5-year unadjusted cumulative incidence of death-censored graft loss from a cohort matched on survival time before GLP-1 receptor agonist initiation was 6·0% for GLP-1 receptor agonist users and 10·7% for non-users (Gray's test p=0·004). The 5-year unadjusted cumulative incidence for mortality from a cohort matched on survival time before GLP-1 receptor agonist initiation was 17·0% for GLP-1 receptor agonist users and 25·8% for non-users (log-rank p=0·0006). The 5-year unadjusted cumulative incidence for mortality was 13·5% for GLP-1 receptor agonist users and 19·9% for non-users (log-rank p<0·0001). GLP-1 receptor agonist use was associated with a 49% lower incidence of death-censored graft loss (adjusted subhazard ratio [aSHR] 0·51, 95% CI 0·36-0·71; p=0·0001) and 31% lower mortality (adjusted hazard ratio [aHR] 0·69, 95% CI 0·55-0·86; p=0·001). Inferences were robust when matched on survival time (death-censored graft loss aSHR 0·53, 95% CI 0·37-0·75; p=0·0005; mortality aHR 0·70, 95% CI 0·55-0·88; p=0·003). Safety endpoints were rare and not associated with GLP-1 receptor agonists, with the exception of diabetic retinopathy (aHR 1·49, 1·11-2·00; p=0·008). INTERPRETATION: GLP-1 receptor agonists were associated with better graft and patient survival. Clinical trials are needed to confirm these findings. FUNDING: National Institutes of Health.