A systematic review and participant-level meta-analysis found little association of retinal microvascular caliber and reduced kidney function.

Lye, Weng Kit; Paterson, Euan; Patterson, Christopher C; Maxwell, Alexander P; Binte Mohammed Abdul, Riswana Banu; Tai, E Shyong; Cheng, Ching Yu; Kayama, Takamasa; Yamashita, Hidetoshi; Sarnak, Mark; Shlipak, Michael; Matsushita, Kunihiro; Mutlu, Unal; Ikram, Mohammad A; Klaver, Caroline; Kifley, Annette; Mitchell, Paul; Myers, Chelsea; Klein, Barbara E; Klein, Ronald; Wong, Tien Y; Sabanayagam, Charumathi; McKay, Gareth J

Previously, variation in retinal vascular caliber has been reported in association with chronic kidney disease (CKD) but findings remain inconsistent. To help clarify this we conducted individual participant data meta-analysis and aggregate data meta-analysis on summary estimates to evaluate cross-sectional associations between retinal vascular caliber and CKD. A systematic review was performed using Medline and EMBASE for articles published until October 2018. The aggregate analysis used a two-stage approach combining summary estimates from eleven studies (44,803 patients) while the individual participant analysis used a one-stage approach combining raw data from nine studies (33,222 patients). CKD stages 3-5 was defined as an estimated glomerular filtration rate under 60 mL/min/1.73m2. Retinal arteriolar and venular caliber (central retinal arteriolar and venular equivalent) were assessed from retinal photographs using computer-assisted methods. Logistic regression estimated relative risk of CKD stages 3-5 associated with a 20 μm decrease (approximately one standard deviation) in central retinal arteriolar and venular equivalent. Prevalence of CKD stages 3-5 was 11.2 % of 33,222 and 11.3 % of 44,803 patients in the individual participant and aggregate data analysis, respectively. No significant associations were detected in adjusted analyses between central retinal arteriolar and venular equivalent and CKD stages 3-5 in the aggregate analysis for central retinal arteriolar relative risk (0.98, 95% confidence interval 0.94-1.03); venular equivalent (0.99, 0.95- 1.04) or individual participant central retinal arteriolar (0.99, 0.95-1.04) or venular equivalent (1.01, 0.97-1.05). Thus, meta-analysis provided little evidence to suggest that cross sectional direct measurements of retinal vascular caliber was associated with CKD stages 3-5 in the general population. Hence, meta-analyses of longitudinal studies evaluating the association between retinal parameters and CKD stages 3-5 may be warranted.