Evidence suggests that leukocyte adhesion and leukostasis play an important role in the pathogenesis of diabetic retinopathy (Lutty et al., 1997
; Barouch et al., 2000
; McLeod et al., 1995
; Schroder et al., 1991
). Leukostasis results not only in capillary nonperfusion, but also leads to endothelial damage and corresponding abnormal autoregulation of blood flow, as well as breakdown of the blood retinal barrier, leading to macular edema in diabetes (Antcliff & Marshall, 1999
) and other diseases (Miyamoto et al., 1996
; Hatchell et al., 1994
). Although confocal microscopes have been used to visualize rolling and sticking leukocytes in the conjuctiva (Kirveskari et al., 2001
), no objective, non-invasive method to measure long-term hemodynamics in human retina have been demonstrated.
The purpose of this study was to develop and demonstrate effective non-invasive ways to determine the effect of the cardiac cycle on parafoveal capillary leukocyte velocity. Better control for changes in blood flow due to pulse would improve the ability to study alterations in leukocyte transport caused by disease and also to evaluate the benefits of pharmaceutical agents used to treat these diseases.
The scanning laser ophthalmoscope (SLO) has established itself as a useful imaging modality for measuring real-time hemodynamics in human retina. However, all reported direct visualizations of blood flow have employed fluorescent contrast agents to improve signal to noise in the images. Real-time SLO imaging combined with traditional fluorescein injections have been used to visualize and quantify blood flow in human retina (Ohnishi et al., 1994
; Wolf et al., 1991
; Yang et al., 1997
). Another method employed fluorescein-labeled autologous leukocytes that were reinjected into the blood stream (Paques et al., 2000
) and visualized with an SLO. Although the fluorescence-based measures are accurate, they are invasive, and the time span for visualization of the leukocytes is limited.
Contrast agents have typically been required to visualize human retinal hemodynamics in the SLO because of limited signal-to-noise in the retinal images, making the smallest features difficult to see. Signal to noise is reduced for two reasons. First, the retinal vasculature is embedded in the retina’s thick, multilayered structure. As such, contrast of the images is reduced because light returning from the retina is comprised of scattered light from multiple structures, not just the vasculature. Second, ocular aberrations of the cornea and lens limit the sharpness (hence contrast) of the retinal images that are recorded.
Adaptive optics (AO) is a technique used in ophthalmoscopy to obtain microscopic access to the living human retina (Liang et al., 1997
). By integrating AO into the SLO imaging modality, video-rate ophthalmoscopy on a microscopic scale is possible in living human eyes (Roorda et al., 2002
; Zhang et al., 2006
). Correcting the aberrations with the AOSLO helps to overcome signal-to-noise limits of retinal images in two ways. First, the AO-corrected image is sharper because of its improved resolution and higher contrast. Second, the AO-corrected confocal optical section becomes narrower and the detected light is limited to the plane containing the vessels of interest (Romero-Borja et al., 2005
). If the subject has normal and otherwise clear optical media, AO makes it possible to image the retina near the diffraction-limit (Zhang & Roorda, 2006
). Although the AOSLO system is conducive to fluorescence imaging (Gray et al., 2008
) and has shown superb dynamic recordings of flow in the terminal ring of capillaries around the fovea of monkeys (Gray et al., 2006
), it also facilitates direct visualization of parafoveal capillary leukocyte dynamics without
fluorescent contrast agents (Martin & Roorda, 2005
). Obviating the need for extrinsic contrast agents makes long term and repeated measures of capillary blood flow possible with AOSLO, and that may be advantageous for tracking response to therapies, for example.
Prior measurements with AOSLO demonstrated that parafoveal capillary leukocyte velocity is variable in normal subjects and we posited that the variability is likely due to pulsatile changes in velocity (Martin & Roorda, 2005
). Observations of pulsatile blood flow in the retina are not new and many techniques have been used to observe or quantify pulsatility. Objective methods include the use of scanning laser Doppler flowmetry and velocimetry (Sullivan et al., 1999
; Grunwald et al., 1986
; Riva et al., 1992
), color Doppler OCT (Yazdanfar et al., 2003
), spectral domain Doppler OCT (Wehbe et al., 2007
), and direct imaging for brief periods (Nelson et al., 2006
). These techniques, however, have to date been unable to record pulsatility in the smallest retinal microvessels and have instead focused on small arteries and veins or optic nerve vasculature. The only demonstrated method that estimates microvascular pulsatility in the parafoveal region employs the blue field entoptic phenomenon (Riva & Petrig, 1980
), which is a subjective technique. In the blue-field entoptic phenomenon method, the subject matches the velocity and pulsatility that they observe entoptically with a simulated flow pattern on a computer screen. It should be mentioned that measurements of pulsatility using SLO with fluorescent contrast agents should, in principle, be possible, but none have been reported to date.