Abstract
Purpose :
Studies have shown that intravitreal anti-VEGF can reduce plasma free-VEGF. However, the literature is sparse on the safety of long-term intravitreal anti-VEGF injections on the body, especially the kidney. We assessed the risk of developing chronic kidney disease or end-stage renal disease in patients receiving multiple anti-VEGF injections.
Methods :
A commercial and Medicare Advantage medical claims database was used to identify all patients who were given an intravitreal anti-VEGF agent. The number of injections was then totaled for the proceeding 4 years following the initial injection. Exclusion occurred if patients at the time of the first injection were <18 years old, had <2 years of data in the plan or had any history of kidney disease (defined by either ICD code or estimated glomerular filtration rate). Patients were then divided into quartiles based on number of injections to compare the highest and lowest quartiles. Subanalysis was also done on those with individual disease states: neovascular AMD, macular edema due to RVO (MERVO), or diabetic retinopathy (DR). A weighted multivariable logistic regression analysis compared the odds of being diagnosed with chronic or end-stage kidney disease. A second time updating analysis, which did not require a specific time in the plan, used Cox proportional regression to assess an every 5-injection increase in the hazard of developing kidney disease.
Results :
13,895 patients in the lowest quartile (representing 1-3 injections) were compared to13881 in the highest quartile (≥21 injections). The nAMD had 6228 (1-4 injections) and 6172 (≥24); MERVO had 573 (1-2) and 580 (≥12); and DR had 523 (1-2) and 517(≥13) in the lowest and highest quartiles, respectively. Weighted analysis showed no association overall (OR=1.00, 95% CI:0.98-1.01, p=0.58) or in nAMD (OR=0.99, 95% CI:0.97-1.02, p=0.48), decreased odds in MERVO (OR=0.93, 95% CI:0.86-0.99, p=0.04) and increased odds in DR (OR=1.12, 95% CI:1.03-1.22, p=0.01) for the highest quartile. Time updating Cox analysis showed no association overall or for any subgroup (HR=0.97-1.01, p>0.27 for all comparisons).
Conclusions :
While logistic regression showed some association of increased number of injections with kidney disease, time updating cox analysis was unable to confirm these findings, suggesting time was a confounder in the logistic regression analysis.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.