Main

Retinopathy of prematurity (ROP), previously termed "retrolental fibroplasia," is a serious vasoproliferative disorder of the retina(1) affecting premature infants. Although many infants have complete resolution of their ROP, some infants progress to develop myopia, poor visual acuity, and blindness(2). With the past and present improvements in neonatal intensive care, the number of extremely low birth weight infants surviving has increased dramatically, resulting in a new generation of children severely visually impaired or blind as a result of ROP(3,4). Their number will probably rise as the neonatal care is expected to improve further in the future. In spite of the advances in neonatal medicine in recent years, no measures are available yet to prevent or to reduce the severity of ROP. Once threshold ROP is established, cryotherapy(5) or laser surgery(6–8) are used to treat the more severe forms of ROP that lead to unfavorable visual outcomes of blindness or severe visual impairment. Even with cryotherapy, 47% of these premature babies had unfavorable visual outcome at 5½ years of age(2).

ROP has a multifactorial etiology. The leading risk factors are birth weight, gestational age, and the number of days on supplemental oxygen. Prolonged mechanical ventilation is associated with bronchopulmonary dysplasia, a severe debilitating chronic lung disease of tiny babies. Corticosteroids are used during the postnatal period in premature babies in two different regimens: the "early" (≤2 wk of age) and "late" (≥2 wk of age) regimen(9–16) for prevention of chronic lung disease (CLD) and for weaning from mechanical ventilation and oxygen supplementation, respectively.

La Motte(17) speculated in 1952 on the potential beneficial effect of adrenal corticotropin hormone on retrolental fibroplasia based on the observation that hyperadrenalism in the post-operative period inhibits the growth of capillaries. In recent years, several clinical studies attempted to assess the effect of postnatal corticosteroids on ROP. The conclusions of these studies are varied(9,11,14–15,17–26). Some have found that use of corticosteroids is associated with an increase in severity of ROP, but others found a protective or no effect at all. The effect of steroids on ROP is controversial as there are no controlled prospective randomized human studies to assess this effect. Recent reports describe a protective effect of antenatal corticosteroids on the development of ROP(27–29). The purpose of this study was to assess the effects of dexamethasone in a mouse model of retinopathy and to test the hypothesis that early use of postnatal corticosteroids during the "critical period" of injury, defined as the period of oxygen exposure, will protect against the development of oxygen-induced retinopathy (OIR) in the mouse.

METHODS

Animal model and dexamethsone experiments. The protocol was approved by the Georgetown University Animal Care and Use Committee. C57BL6 mice were obtained from Taconic Laboratories (Germantown, NY). Mice exposed to oxygen treatment were placed with their nursing dam in an infant incubator (Ohmeda, Inc., Columbia, MD) from PN7 (postnatal d 7) through PN12 as previously described(30). The oxygen flow was adjusted to keep the concentration in the incubator at 75 ± 2% FiO2. Oxygen concentration was measured using a Hudson Oxygen Analyzer (Hudson Ventronics, Temecula, CA) and was checked at least twice daily during the period of exposure. Animals assigned to dexamethasone treatment group were given a single dose of 0.5 mg/kg/d of dexamethasone (American Reagent Laboratories, Inc., Shirley, NY) s.c. in the nape of the neck for 5 d beginning at PN7. This dose is comparable to that used in clinical settings for prevention of CLD or weaning from mechanical ventilation. On PN12 animals were removed from the oxygen and placed in room air. From PN12 through PN17-20 all animal groups were kept in room air and then killed by a lethal intraperitoneal injection of sodium pentobarbital (Abbott Laboratories, North Chicago, IL). The greatest neovascular response occurred in this model from PN17 to 21(30), and thus all animals were killed at PN17-20 in these studies. Animal weights were recorded on PN7, PN12, and PN17-20. A log of animal death was kept throughout the experiment.

Fluorescein dextran perfusion of the retinal blood vessels. To study the retinal vascular pattern, systemic perfusion was performed(31) using high molecular weight (MW = 2 000 000) fluorescein-conjugated dextran (Sigma Chemical Co., St. Louis, MO) in PBS (GIBCO, Grand Island, NY). Briefly, animals were given a lethal dose of sodium pentobarbital and a median sternotomy was performed. The left ventricle was identified and perfused with fluorescein-conjugated dextran (50 mg/mL in 4% PBS) using a 1-mL tuberculin syringe with a 27-gauge needle. Eyes were then enucleated and placed in 4% paraformaldehyde (Sigma Chemical Co.) in PBS for 4 to 24 h. Using a dissecting microscope, the retina was removed and a flat mount was prepared by making radial cuts. A coverslip was applied over the retinae after placing a drop of 2% gelatin (Sigma Chemical Co.). The edge of the coverslip was sealed with transparent nail polish. The scoring of retinal whole mounts was performed using fluorescent microscopy. Each retina was scored by at least two independent observers in a masked fashion using the retinopathy scoring system(32) as shown in Table 1 and the average retinopathy score for each animal was used for statistical analysis.

Table 1 Retinopathy scoring system
Full size table

Periodic acid-Schiff (PAS) stain of retinal sections. Mice were killed on PN17-20 using sodium pentobarbital. After midline sternotomy was performed, the left ventricle was injected with 4% paraformaldehyde in PBS. The eyes were enucleated, placed immediately in optimum cutting temperature embedding compound (Sakura Fine Tek, Inc., Torrence, CA), and frozen at -70°C. Serial sections (8-9 microns thick) over a minimum of 450 microns were cut in a sagittal plane through the cornea, parallel to the optic disc. Tissue sections were stained with periodic acid-Schiff (PAS) reagent and hematoxylin(33). Multiple sections from individual eyes were scored in a masked fashion using light microscopy counting all nuclei extending beyond the inner limiting membrane into the vitreous as previously described(30). A minimum of six sections at least 50 microns apart were evaluated and counted per eye and then averaged. The average number of neovascular nuclei for each eye was used in the statistical analysis.

Animal growth and death. To assess the effect of dexamethasone and 5 d of oxygen exposure on animal well being, a log of animal death was kept throughout the experiment. Weights were recorded on PN7, PN12, and PN17-20 using a laboratory balance.

Statistical analyses. Each retina was scored by two independent observers. The scores were averaged for the two eyes and each animal had one retinopathy score, which was used for statistical analysis. Analysis of variance using the Kruskal-Wallis test was performed to test for differences among the various treatment groups. Mann-Whitney tests were used to compare the total retinopathy scores and the retinopathy subscores between groups. t tests were used to compare nuclei count and animal weight. Statistical significance was defined as p < 0.05. Interobserver variability for the retinal scoring system was assessed by using the Kendall's tau-b correlation coefficient test.

RESULTS

Total retinopathy scores and subscores. Dexamethasone significantly decreased the total retinopathy scores as shown in Figure 1. Control animals (n = 11) had a median total retinopathy score [median (25th, 75th quartile)] of 1 (0.5, 1.5). Animals treated with dexamethasone only (n = 17) had a total retinopathy score of 0.5 (0,1). Oxygen exposed animals (n = 14) had a median total retinopathy score of 9(6,10) versus oxygen and dexamethasone treated animals (n = 12) with median total retinopathy score of 5(4,6). The correlation between the observers' retinopathy scores was significant at the 0.01 level by the Kendall's tau-b (Fig. 2) demonstrating concordance of the independently obtained scores.

Figure 1
figure 1

Total retinopathy scores - control and oxygen treated animals with and without dexamethasone. Animals exposed to 75% oxygen had median retinopathy score of 9. Animals exposed to oxygen and dexamethasone had median retinopathy score of 5 (p < 0.001). There is no difference (p = 0.31) in retinopathy scores between control animals and those treated only with dexamethasone. Data are shown as median ± 25th-75th quartile. Oxygen exposed animals are indicated by (○) and control animals by (·).

Full size image
Figure 2
figure 2

Kendall's tau-b interobserver correlation. There was significant correlation of the two sets of scores at the 0.01 level.

Full size image

Animals treated with dexamethasone and oxygen compared with animals treated with oxygen only had a significant reduction in the subscores of blood vessel tufts, extra retinal neovascularization, and blood vessel tortuosity (p < 0.001) as shown in Table 2. Figure 3 shows retinal whole mounts from the various groups. Animals treated with dexamethasone, compared with control group, did not show any difference in the total retinal scores or subscores (Fig. 1, Table 2), hence dexamethasone does not have an effect on the normal retinal development as measured by the retinopathy scoring system.

Table 2 Subscores for the categories of retinopathy
Full size table
Figure 3
figure 3

Retinal whole mounts from various treatment groups. Note the decrease in severity of OIR, as measured by decrease in the number of tufts, severity of extra retinal neovascularization, and blood vessel tortuosity, in the animals treated with dexamethasone and oxygen (B) compared with oxygen treated animals (A). (A) Hyperoxia group: Retina from oxygen treated animal shows an abundance of blood vessel tufts, extraretinal neovascularization, central loss of blood vessels or vasoconstriction, and severe blood vessel tortuosity. (B) Oxygen and dexamethasone group: Retina from oxygen and dexamethasone treated animal, compared with (A) shows significant reduction in the number of vascular tufts, extra retinal neovascularization, and blood vessel tortuosity. (C) Control group: The vessels have a fine pattern throughout. Note the absence of blood vessel tufts, extra retinal neovascularization, central vasoconstriction, and blood vessel tortuosity.

Full size image

Retinal sections. To corroborate on the finding of extra-retinal neovascularization on flat retinal mounts, retinal sections were performed(30). Control and dexamethasone only group had an average of 3.1 ± 1.1 and 4.4 ± 2.7 (average ± SD) nuclei per section, respectively (p = 0.3). Oxygen compared with oxygen and dexamethasone treated animals had an average of 43.4 ± 13.1 and 27.3 ± 17.1 neovascular nuclei per section, respectively (p = 0.04), as shown in Figure 4.

Figure 4
figure 4

Effect of dexamethasone on neovascular nuclei. Oxygen treated animals had 43.4 ± 13 (mean ± SD) nuclei extending beyond the inner limiting membrane. Oxygen and dexamethasone treated animals had 27.3 ± 17 nuclei (p = 0.04). There is no difference between control and dexamethasone treated animals (p = 0.3). Oxygen treatment appears as (○) and control treatment as (·).

Full size image

Animal growth and death. Animal weight and death was monitored to assess the effect of 5 d of oxygen and dexamethasone exposure on animal health. Weights were recorded on PN7, PN12, and PN17-20. Animals exposed to dexamethasone had a slower rate of growth while exposed to the drug (Table 3). This finding was previously described in literature(13,15–16,25,34–36). After discontinuation of the drug on PN12, growth in the dexamethasone treated animals was similar to the control animals as assessed by average weight gain per day from PN12 until the day the animals were killed. Animals still had statistically lower weights on PN17-20. There were no deaths in the control group and one death in each of the other groups of animals. The deaths were excluded from the weight analysis.

Table 3 Animal weight
Full size table

DISCUSSION

The results demonstrate a clear beneficial effect of dexamethasone, when given concurrently with oxygen exposure, on OIR in the mouse. There was a significant reduction in the total retinopathy scores in the dexamethasone and oxygen treated animals when compared with oxygen treated animals. The specific subscore categories showing improvement were blood vessels tufts, extra retinal neovascularization, and blood vessel tortuosity. Dexamethasone and oxygen treated animals have a significantly lower number of neovascular nuclei on PAS-stained retinal sections when compared with animals treated with oxygen only. Although a beneficial effect was observed in the development of retinopathy in the mice, dexamethasone did cause growth impairment.

The mouse model of OIR is not a true representation of human ROP. Hyperoxic injury followed by retinal hypoxia produces an aggressive retinopathy in the mouse. Human ROP is characterized by an initial delay in vascular development and vasoconstriction/vaso-obliteration with subsequent hypoxia-driven retinal neovascularization. The dexamethasone used in these mice pups was given during the exposure to 75% oxygen and retinopathy was evaluated at the peak of vasoproliferation (PN17-20). Seventy-five percent oxygen is not routinely clinically used; the lowest amount of supplemental oxygen (if any) to maintain arterial oxygen tension in a normal range is used for premature infants. In addition, mice have retinal vasculature that normally matures ex utero at term; human infants do not. Premature infants have retinal vasculature that must mature in a developmentally abnormal (ex utero as opposed to in utero) environment where arterial oxygen tensions are higher than in utero. ROP is characterized by neovascularization at the developing vessel front. Mice develop retinal neovascular blood vessel tufts at the end of the area of central vasoconstriction. This model has been used to study down-regulation of vascular endothelial growth factor by hyperoxia(37) and modulation of vasoconstriction with calcium channel blockade(32), both of which target the injury produced in this model by 75% oxygen exposure for 5 d. The mouse model has been used extensively to study retinal neovascularization, which occurs after room air recovery of the animals(38–43). Injury and healing occur in both the mouse model of OIR and ROP, but there are species differences.

Although the precise mechanism by which dexamethasone exerts its protective effect is unknown, we speculate that the protective effect may by mediated, at least in part, by reduction in inflammation and/or injury caused by 75% oxygen exposure. Oxygen promotes synthesis and secretion of TNF(44–47). TNF is a macrophage, monocyte, or endothelial cell-derived proinflammatory polypeptide cytokine(48). Jensen(47) showed that the lungs of mice exposed to high-dose oxygen (greater than 95%) for 3 d demonstrated an increased expression of the gene for TNF. Daily treatment of these animals with anti-TNF antibodies improved their survival. These findings suggest that oxygen exposure induces TNF, which explains part of the toxicity of oxygen. O'Brien-Lander(45) showed that hyperoxia exposure enhances TNF production by posttranscriptional mechanism in LPS-stimulated human alveolar macrophages. Many investigators have documented that dexamethasone inhibits TNF production(49–53). Joyce(49) showed that dexamethasone suppresses both spontaneous and LPS-stimulated release of soluble forms of receptors for TNF from monocytes and suppresses the release of bioactive TNF from peripheral blood mononuclear cells. Brenner(50) used astrocytes derived from fetal rat brain and triggered by mycoplasmas to produce TNF. Preincubation of these cells with dexamethasone markedly inhibited the secretion of TNF. Waage(51) demonstrated that the production of TNF by LPS-stimulated human monocytes was significantly inhibited in a time and a dose-dependent fashion by dexamethasone.

TNF is found in the retina(54–57) and has been implicated in proliferative vitreoretinopathy(56,57). Modulation of TNF by dexamethasone may explain the improvement in retinopathy observed in the mouse model. Dexamethasone may inhibit inflammation via a TNF mediated macrophage infiltration thus protecting fragile capillaries from injury during the hyperoxic exposure.

Intravitreal administration of steroids has been shown to inhibit preretinal neovascularization in a pig model(58) and subretinal neovascularization in a primate model(59). The mechanism of inhibition of blood vessel proliferation is unclear, but is postulated that steroids have anti-inflammatory properties. Inhibition of macrophage infiltration to an area of injury has been proposed as a protective mechanism of action(58,59).

Dexamethasone may stabilize retinal capillary endothelial cells making them less susceptible to the injurious effects of 75% oxygen in the mouse model. Dexamethasone has been shown to stabilize human umbilical vein endothelium(60). Perhaps the dexamethasone treated animals are protected from the injury of the 75% oxygen more than their control counter-parts.

Timing of dexamethasone dosing may be critically important as pointed out by Ehrenkrantz(24). Dexamethasone administered to infants earlier (median d 23 versus 38) for pulmonary indications showed significantly lower rates of cryosurgery(21). The timing of the dexamethasone used in our experiments was most likely equivalent to early postnatal dexamethasone exposure. The developmental stage of the blood vessels in the newborn mouse retina approximates 4 to 5 mo gestation in the preterm infant(61). Antenatal corticosteroids have recently been shown to protect against the development of severe ROP(27–29). It has been speculated that antenatal corticosteroids may accelerate the maturation of the retinal vasculature, thus placing the infant at lower risk of developing serious ROP(29). The steroids may decrease the response of the developing blood vessels to the injurious extra uterine environment perhaps through TNF-mediated mechanisms. Dexamethasone in the mice may protect the retinal vasculature by an unknown or undescribed mechanism rendering by an unknown or undescribed mechanism rendering them less susceptible to retinal neovascularization when removed from 75% oxygen on d 12 of the experiment. The dexamethasone used in these animal experiments may be acting by a similar mechanism that renders antenatal steroids protective.

In conclusion, the retinopathy scoring system and quantification of neovascular nuclei on retinal sections show a beneficial effect of dexamethasone on OIR. Further studies are needed to establish the exact mechanism of action by which dexamethasone exerts its protective effect and to establish the best timing for dexamethasone exposure.