Validity of a novel optical coherence tomography angiography flow index in a cohort of primary open angle glaucoma

Abstract Background Vascular mechanisms are implicated in many ocular diseases. Therefore, different vascular imaging modalities are used in management of such conditions. Optical coherence tomography angiography (OCTA) has high spatial resolution and segmentable 3D volumetric sampling enabling isolation of retinal and peripapillary vascular beds. However, OCTA only indirectly derives quantitative flow data i.e. velocimetry through methods and algorithms liable to limitations like signal saturation. This study introduces and validates novel mathematical OCTA flow indices that may compensate for some OCTA velocimetric limitations. Methods Thirty-seven eyes of 23 POAG patients were included. Each underwent baseline and follow-up assessment one month thereafter. Assessment comprised full ophthalmological examination, intraocular pressure (IOP), systemic arterial blood pressure (SABP) and OCTA macula and ONH. Angiograms were processed using ImageJ to calculate OCTA intensity-based flow indices (FIOs), for superficial vascular plexus (SVP), deep vascular plexus (DVP) and optic nerve head vascular plexus (ONH-RPC), i.e. SFIO, DFIO and ONHFIO respectively. Mean ocular perfusion pressure (MOPP) was calculated using IOP and SABP. OCTA vascular densities (VD) and MOPP were used to calculate three respective mathematical flow indices (FIMs) for SVP, DVP and ONH-RPC, based on Hagen-Poiseuille law, i.e. SFIM, DFIM, ONHFIM respectively. Pearson test was used for correlation between the two sets of indices. Intraclass correlation coefficient (ICC) was tested for baseline and follow-up values for each index. Results There was positive correlation between the three FIMs and their respective FIOs at baseline and follow-up ranging between high and moderate. Correlation coefficients (CCs) were 0.773 and 0.609 for SFIM and SFIO P-value < 0.001, 0.829 and 0.624 for DFIM and DFIO P-value < 0.001 and 0.516 and 0.737 for ONHFIM P-value = 0.001 for baseline and follow-up respectively. ICCs were 0.772 P-value < 0.001, 0.328 P-value = 0.022 and 0.888 P-value < 0.001 for SFIM, DFIM and ONHFIM respectively. For SFIO, DFIO and ONHFIO, ICCs were 0.420 P-value = 0.004, 0.079 P-value = 0.320 and 0.833 P-value < 0.001 respectively. Conclusion The novel FIMs are reliable alternatives to FIOs and may compensate for OCTA signal saturation in extremes of MOPP. SFIM and ONHFIM showed high ICCs with excellent reliability. While DFIM demonstrated low ICC indicating poor reliability, it still performed better than its corresponding DFIO