Relationship between Retinal Vessels and OCT-Derived Retinal Neural Parameters

ObjectiveTo investigate structural relationships between retinal vasculometry derived from color fundus photography (CFP) and neural parameters obtained from Optical Coherence Tomography (OCT) scans and validate their causal relationships. DesignCross-sectional study ParticipantsParticipants with fundus photographs data and OCT data in the UK Biobank cohort study MethodsWe used the Retina-based Microvascular Health Assessment System (RMHAS) to extract retinal vascular measurements in the 6*6mm area centered on the macular region. OCT parameters were available from the UK Biobank. First, pairwise correlations between individual retinal layers and vascular parameters were investigated. Canonical correlation analysis (CCA) was then used to examine associations between sets of variables. Lastly, bidirectional two-sample Mendelian randomization was employed to investigate potential causal relationships. Main Outcome MeasuresMeasurements of retinal vascular network and neural layers ResultsData from 67,918 eyes of 43,029 participants were included. The Ganglion Cell-Inner Plexiform Layer (GC-IPL) thickness showed the strongest correlations with vascular Density and Complexity (r=0.199 for arterial Vessel Area Density and r=0.175 for Number of Segments). The Inner Nuclear Layer (INL) thickness showed positive correlations with Width (r=0.122) and Vessel Area Density (artery) (r=0.127). Mendelian randomization analysis indicated bidirectional causal relationships between retinal vascular features and layer thicknesses. Genetically predicted higher Vessel Density was associated with increased thickness across various retinal layers, with the strongest effect on Inner Segment/Outer Segment + Photoreceptor Segment thickness (standardized effect size 1.50, p<0.001). Genetically predicted increases in retinal layer thicknesses, particularly the Outer Plexiform Layer, were linked to higher Vessel Density (standardized effect size 0.45, p=0.002) and Fractal Dimension (standardized effect size 0.48, p<0.001). ConclusionsThe positive associations of macular thickness with vascular Density and Caliber measurements were mainly attributable to their associations with GC-IPL and INL. Multidimensional relationships revealed by CCA revealed a complementary nature between the two sets of parameters, highlighting their value as a composite biomarker. Mendelian Randomization uncovered a bidirectional causal relationship that should provide insights into novel therapeutic approaches targeting both vascular and neuronal components.