Abstract
Purpose: :
This study applies recently developed retinal lamina–specific MRI technologies to evaluate early stage diabetic retinopathy (DR) in streptozotocin (STZ) treated rats. We hypothesized that MRI can be used to resolve retinal lamina–specific structural and functional changes associated with DR. We first determined the thickness of distinct retinal layers using high resolution anatomic MRI. We then applied BOLD fMRI to determine BOLD fMRI responses to hyperoxia and hypercapnia in diabetic and non–diabetic control retinas.
Methods: :
Age–matched male rats were injected i.v. with STZ (100 mg/kg, n=6) or vehicle (n=8). Imaging was performed at 3.5 months after STZ injection. Anatomical MRI was acquired TR=100ms, TE=4ms, slice thickness=0.8mm, FOV=8x8mm, matrix=128x128. BOLD fMRI was acquired using spin–echo EPI with diffusion weighting to suppress the vitreous signal, with similar parameters except TR=1s, TE=20ms, FOV=1.1x1.1cm. Physiological challenges used 100%O2 (hyperoxia) or 5%CO2 (hypercapnia, 21%O2). BOLD %–change maps, BOLD % changes and number of activated pixels were analyzed.
Results: :
Anatomical images resolved three robust layers in the retina of both diabetic and control rats. In controls, the thicknesses of the inner, middle and outer layers were 160±24, 84±16, and 93±16µm, respectively. In diabetic group, respective thicknesses were 170±23, 72±12, and 132±34µm. The outer layer thickness in diabetic rats was statistically greater than in controls. Robust lamina–specific BOLD fMRI associated with physiological stimuli were also observed. Comparing BOLD % change in the outer and inner strip, hyperoxia–induced significantly larger changes in the outer strip while hypercapnia–induced greater change in the inner strip. The physiologically induced % changes in different layers were however not statistically different between diabetic and control group (P>0.05). The numbers of activated pixels in all layers were diminished in diabetes relative to controls by 42% (P<0.01) for hyperoxia and by 33% (P<0.01) for hypercapnia. These results suggest that neurovascular coupling is perturbed in early phase DR.
Conclusions: :
MRI thus has the potential to be used to non–invasively stage the progression of diabetic retinopathy and to provide powerful insights into how DR independently affects structures, functions, retinal vasculatures and the neural tissues they subserve. Imaging at earlier and multiple time points and correlation with electroretinography and histology are being investigated.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • diabetes • retina