Dissecting the Neurovascular Unit in Human Diabetic Retinal Disease Free

In this issue of Diabetes, Yang et al. (1) analyze the relationships between the microvascular and neurodegenerative aspects of diabetic retinal disease (DRD) in eyes of donors with diabetes, the varying degrees of clinically evident microvascular lesions, and the eyes of control donors without diabetes. Briefly, they conclude that the microvascular and neurodegenerative aspects of disease are not necessarily connected spatially.

DRD has been considered the prototypic “microvascular disease” of diabetes, but substantial evidence now reveals alterations in neurons, glial cells, and microglial cell function and structure at least as early as the onset of microvascular permeability or capillary dropout (2,3). The current report is important because studies conducted six decades ago focused on microvascular lesions following trypsin digestion of the neurosensory retina (4) or on neurodegenerative lesions (5,6), with little understanding of how the normal neurovascular unit progressively disintegrates in diabetes (7) and without the advantage of contemporary immunohistochemical methods. Clinical-pathological comparisons by Bek (8) and fluorescein angiographic–optical coherence tomographic studies by Murakami and colleagues (9) have revealed a strong regional correspondence between capillary nonperfusion and neuroretinal cell loss, at least in advanced DRD. Quantitative analyses of the human retinal neurovascular unit in normal and diabetic retinas are limited. Balaratnasingam and colleagues have developed standardized methods to assess functional capillary perfusion in human retinas compared with in vivo optical coherence tomography angiography (10,11). Hence, the study of Yang et al. is well timed to address knowledge gaps in the field.

Yang et al. conducted detailed histopathologic analyses of postmortem human retinas from 14 donors: 10 with diabetes but no retinal vascular lesions, 1 with overt diabetic vascular lesions, and 3 without diabetes (control). The salient findings are as follows: 1) microvascular defects first appear in the deep capillary plexus that nourishes the outer retina when no other vascular lesions are found and 2) degeneration of neural cell layers across the retina occurs without definite connection to microvascular lesions.

The very early defects in the deep retinal vascular plexus are consistent with similar findings using optical coherence tomography angiography (12), a method that can resolve perfusion defects in capillaries that are not detectable by conventional fluorescein angiography. The widespread inner nuclear layer cell loss is consistent with findings by Sohn et al. (3), but the apparent dissociation of vascular nonperfusion from neuronal cell loss is not predicted, although it suggests that neuronal damage may be at least partly independent of microvascular lesions. The importance of these findings is that clinically useful structural biomarkers of disease may be developed.

The study was conducted by experts in clinical ophthalmology, neuroimaging, and retinal vascular biology whose combined strengths enabled the findings. Inherent limitations of postmortem studies are pointed out by the authors, including the long death-to-fixation times shown in Table 1. Several other points merit consideration.

A quantitative approach requires statistical analysis, which is difficult given the low number of individuals. The authors attempted to solve the problem by including 18 sections from each donor. A more appropriate approach would have been to use two-level or nested ANOVA with the sections as the first level and the individual as the second level. However, this would not eliminate the challenge of having too few individuals within each group. Level two in the nested ANOVA would contain the information but would be underpowered to allow conclusions to be made. In Fig. 3, the red signatures represent one observation from the patient with DRD, but, despite this, the data is presented in statistical terms. The main message of the article is the finding of a 7% difference in inner neuron layer cells between two groups with three and four observations, so it will require substantiation. Moreover, it would be helpful to know the impact of diabetes on retinal ganglion cell numbers.

In studies of total retinas, there is a lack of information about the details, which is problematic with a disease characterized by localized lesions. This issue is exemplified by the finding of a uniform next-neighbor distance between capillaries, as this parameter varies within individual microcirculatory units even in normal retinas. Thus, one may overlook significant information.

It is difficult to measure vessel diameters in postmortem tissue. Perfused vessels collapse at death when the blood pressure becomes zero, and occluded vessels undergo shrinkage during the histological processing. These changes are individual and are difficult to control.

An outstanding question is how the density of immunoreactivity to type IV collagen was assumed to represent capillary basement membrane, irrespective of coexisting endothelial cells. This density (representing both actual and former capillary cross sections) should be the reference for studies of vessel densities.

The authors distinguish between the superficial and deep retinal capillary plexus, but in the retinal periphery there is only one layer; the macula has three layers, and the peripapillary nerve fiber layer has four layers. It would be useful to know how was this handled in the sections where all these areas were represented.

Notably, does capillary nonperfusion develop due to intralumenal microvascular obstruction and/or does neuroglial cell dysfunction reduce blood flow? From this question, what ocular or systemic factors can stimulate regeneration of the neurovascular unit and its blood flow, as documented in prior case reports (12,13)? Finally, this work is essential to develop an understanding of neurovascular relationships that can lead to a quantitative grading scale for DRD that reflects the comprehensive status of the retina (14,15).