Spatial Pattern and Temporal Evolution of Retinal Oxygenation Response in Oxygen-Induced Retinopathy

Abstract

purpose. To determine the spatial pattern and temporal evolution of the change in retinal partial oxygen pressure (ΔPo ₂) associated with a murine oxygen-induced retinopathy (OIR) model of retinal neovascularization (NV).

methods. On P7, newborn C57BL/6 mice were exposed to 75% oxygen until postnatal day (P)12, followed by recovery in room air until P17 or P34. Control mice remained in room air until P17 or P34. At P17 and P34, functional magnetic resonance imaging (MRI) and a carbogen inhalation challenge was used to measure retinal ΔPo ₂. Retinal avascularity, distance from the optic nerve head to the vascular edge in the peripheral retina, and NV incidence and severity were measured in retinas stained with adenosine diphosphatase (ADPase).

results. In P17 and P34 controls and in P34 OIR animals, retinas were fully vascularized without evidence of NV. In P17 OIR mice, there was a large central retinal capillary-free zone (22% ± 3% of the entire retinal area, mean ± SD) and 4 clockhours (range 1–7) of retinal NV at the border of the peripheral vascular and central acapillary retina in 100% (36/36) of the mice. In P17 OIR mice, retinal ΔPo ₂ over the vascularized far peripheral retina was not significantly (P > 0.05) different from the P17 control but was supernormal (P < 0.05) over the central capillary-free retina. However, no differences (P > 0.05) in retinal ΔPo ₂ were found between the P34 control and OIR groups.

conclusions. A reversible supernormal ΔPo ₂ was found only over the central acapillary retina during the appearance of retinal NV in a mouse OIR model. The present data show the applicability of carbogen-challenge functional MRI to the study of retinal ΔPo ₂ in vivo in eyes that are too small for the use of existing techniques.

Very low birth weight infants at high risk of development of retinopathy of prematurity (ROP), with associated vision loss and blindness, may have several retinal vascular irregularities, including retinal vessel dilation, tortuosity, and neovascularization (NV) at the border between the peripheral vascular and avascular retina. ¹ The pathophysiology associated with such circulatory abnormalities is not well characterized, but impaired hemodynamic function is thought to be linked with these vascular abnormalities. ^{1 2 3 4 5}  

To study retinal NV, several experimental models have been developed. The most common models involve modification of the inspired oxygen level of newborn animals (e.g., oxygen-induced retinopathy [OIR] and variable OIR). ^{6 7 8 9 10 11 12} In the typical OIR model, neonatal C57BL/6 mice are exposed to a constant high (75%) oxygen level between postnatal day (P)7 and P12. ⁷ By P12, the hyperoxic exposure results in the disappearance of existing capillaries in the central retina (although the peripheral retina remains vascularized). Recovery of these animals in room air until P17 allows revascularization of this central avascular portion of the retina with associated marked retinal NV at the border between the central avascular and peripheral vascular retina. ¹³ Another frequently used experimental model of ROP involves exposing newborn rats to a variable oxygen environment. ¹⁴ In this experiment, typically, newborn Sprague-Dawley rats are exposed from P0 to P14 to an environment that alternates between 50% and 10% oxygen every other day (i.e., a 50/10 model). By P14, the variable oxygen-induced attenuation of retinal vessel growth produces a large peripheral avascular region. Between P14 and P20, rats breathe room air, and the peripheral retina becomes more vascularized. By P20, retinal NV is consistently found at the border between the peripheral vascular and avascular retina. Although both OIR and 50/10 variable oxygen models reliably produce retinal NV, the conditions (i.e., type of insult and timing) and retinal regions involved are substantially different. The small size of the eye and presence of a hyaloidal circulation has made the measurement of retinal hemodynamic parameters in newborn rodent models difficult with existing methods. ¹⁵ For this reason, it is not yet known whether any retinal physiologic parameters are different between the OIR and variable oxygen models. 

We have developed a novel functional magnetic resonance imaging (MRI) method for accurately measuring one aspect of retinal physiology: the retinal oxygenation response to a hyperoxic inhalation challenge. ^{16 17 18} In this method, hyperoxia increases vitreous partial oxygen pressure over room air levels (ΔPo ₂). Because oxygen is paramagnetic, this ΔPo ₂ produces an increase in the vitreous signal intensity on a T₁-weighted image. In normal newborn rats, carbogen breathing oxygenates the retina significantly better than pure oxygen breathing. ¹⁹ Carbogen is a gas mixture of carbon dioxide (5%) and oxygen (95%) that has been used clinically instead of 100% oxygen, to minimize the vasoconstrictive effects of pure O₂ on retinal blood flow and oxygenation. Using this acute hyperoxic inhalation challenge as an acute retinal stress test in the 50/10 variable oxygen model, we found the following spatial pattern and temporal evolution of retinal ΔPo ₂: (1) The vascular bed from which the NV develops and peripheral avascular retina had significantly lower ΔPo ₂ than similar retinal regions in vascularized retina in age-matched control rats, and (2) a subnormal panretinal response was found during and after the appearance of retinal NV. ^{3 4} The purpose of the present study was to determine the spatial pattern and temporal evolution of retinal ΔPo ₂ in the mouse OIR model. 

Methods

The animals were treated in accordance with the NIH Guide for the Care and Use of Laboratory Animals and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Animal Model

The newborn mouse OIR model has been described in detail elsewhere. ⁷ Briefly, C57BL/6 dams and their litters are maintained in room air until P7. They are then placed in a modified pediatric incubator where the oxygen level is maintained constant for the next 5 days (until P12). Animals are then allowed to recover in room air (21%) during the next 5 days (until P17). 

Functional MRI Examination

The functional MRI procedure has been described in detail elsewhere. ^{3 4} Briefly, on the day of the examination, urethane-anesthetized animals (0.083 mL of a 36% solution of urethane per 20 g animal weight, administered intraperitoneally [IP] and freshly made daily; Aldrich, Milwaukee, WI) were gently positioned on an MRI-compatible custom-made holder with the nose placed in a plastic nose cone and were allowed to breathe spontaneously. Rectal temperature was continuously monitored to maintain body temperature. MRI data were acquired on a 4.7-T system, using a two-turn surface coil (1.5-cm diameter rat, 1-cm diameter mouse) placed over the eye and a spin-echo imaging sequence (repetition time [TR] 1 second, echo time [TE] 22.7 ms, number of acquisitions [NA] 1, matrix size 128 × 256, slice thickness 1 mm, field of view 16 × 16 mm² (mouse), sweep width 25,000 Hz, 2 minutes/image). A capillary tube (1.5 mm inner diameter) filled with distilled water was used as the external standard. Four sequential 2-minute images were acquired—three control images while the animal breathed room air and one image during carbogen breathing. In each animal, carbogen exposure was started at the same phase-encoding step near the end of the third image. This procedure was followed exactly in every animal. If an accident prevented this timing, the experiment was aborted, the inhalation gas was switched to room air, and the study was begun again after a 5-minute reset time. Carbogen gas was chosen for these studies because it induces a maximum oxygenation response in the retinal circulation (compared with oxygen alone), allowing more robust detection of subtle differences between groups. ¹⁹  

At the end of the carbogen challenge, the animals were returned to room air and removed from the magnet. Blood, obtained from a cut in the carotid artery was collected immediately after a second 2-minute carbogen challenge and analyzed for partial arterial oxygen (P_aO₂) and carbon dioxide (P_aCO₂) pressures and pH. Note that this second inhalation challenge (outside the magnet) is necessary because it is not feasible to routinely obtain an arterial blood sample from inside the magnet (>40 cm away from the magnet opening) in newborn rodents. After the blood collection, the animal was killed with an intracardiac potassium chloride injection, the eyes were enucleated, and the retinas wholemounted. 

The increase in partial oxygen pressure in the vitreous over the room air value (ΔPo ₂) was detected as an increase in the signal intensity on a T₁-weighted image, as previously described. ^{3 16 19} It is important to note that steady state (room air) vitreous oxygen tension cannot be measured using this method, because many factors (e.g., vitreous temperature and protein content) affect the baseline preretinal vitreous water signal and its relaxation properties. In other words, simply obtaining an image of the eye during room air breathing alone cannot be used to measure retinal oxygenation. However, these factors are not likely to change on the short time scale between baseline and carbogen breathing. Thus, their contributions are expected to cancel one another and not contribute to the ΔPo ₂ measurement. 

Data Analysis

To be included in this study, the animal must have demonstrated (1) minimal eye movement during the MRI examination. Movement artifacts (typically seen in the phase-encoding direction) would confound interpretation of the vitreous signal intensity changes produced during the hyperoxic challenge; (2) a nongasping respiratory pattern before and after the MRI examination. If the animal is gasping (which occurred <1% of the time), the anesthetic was probably improperly administered (e.g., not IP). This could produce a change in systemic oxygenation unrelated to the retinal changes; (3) rectal temperatures in the range of 35.5°C to 36.5°C. Preliminary experiments (data not shown) found a strong association between core temperature and P_aCO₂ and P_aO₂ levels. The effect of this correlation on the precision of the measurements was minimized by using a relatively tight range of temperatures; and (4) P_aO₂ higher than 350 mm Hg and P_aCO₂ between 46 and 65 mm Hg during the carbogen challenge. We have found that arterial oxygen levels above 350 mm Hg during a hyperoxic challenge are needed to produce a consistently large preretinal vitreous oxygenation response. ²⁰ The range of acceptable arterial carbon dioxide levels lies within the array of values in the literature that were measured under carbogen breathing conditions. In addition, tight control over the acceptable blood gas range is needed to ensure adequate quality control of each sample. Occasionally, the blood gas machine was not able to read a sample (e.g., due to a clot or excessive air in the capillary tube). In this case, the MRI data were also excluded. These stringent acceptance criteria are necessary to compare critically the retinal oxygenation response in these spontaneously breathing normal and sick animals while minimizing systemic differences. Because such strict criteria were used, only approximately 50% of the animals that started the study were used in the final analysis. Based on our previous experience in rats, n ≥ 5 is sufficient to draw statistical conclusions. In the present study, six P17 control, six OIR, eight P34 control, and seven OIR mice satisfied the inclusion criteria and were used for data analysis and statistical comparisons. 

The ΔPo ₂ parameter image was analyzed as follows: First, the pixel values along a 1-pixel-thick line (in-plane resolution 125 × 63 μm²) drawn at the boundary of the retina/choroid and vitreous were set to 255 (black). The values in another 1-pixel-thick line drawn in the preretinal vitreous next to the black pixels were then extracted. ³ This procedure minimized retinal–choroid pixel values from potentially contaminating (pixel bleed) those used in the final analysis and insured that similar preretinal vitreous space was sampled for each animal. In addition, spatial averaging over the 125 × 63 μm² pixels tends to minimize the contribution from the very local preretinal oxygenation gradients next to the retinal surface. ²¹  

As previously described, from each animal in this study, a map of preretinal vitreous ΔPo ₂ values was extracted and used for further data analysis. ^{3 4 17 22 23} The summary of the retinal ΔPo ₂ results is presented in two formats. Within each group, a plot of the spatial variations in retinal ΔPo ₂ was constructed by first averaging values at each pixel location and then averaging pixels equidistant from the optic nerve head in the superior and inferior directions. Alternatively, a measure of mean retinal ΔPo ₂ in each group was obtained by averaging all pixel values within a region of interest, regardless of spatial location. For clarity, error bars are presented only for the mean response and not on the spatial maps. 

Statistical Analysis

The physiological parameters (i.e., blood gas values, rectal temperatures, blood glucose data) were normally distributed and are presented as mean ± SEM. Comparisons were performed using one-way ANOVA with the Tukey post test. Statistical comparisons of retinal ΔPo ₂ between control and experimental groups were performed using a generalized estimating equation approach. This method performs a general linear regression analysis using all the pixels in each subject and accounts for the within-subject correlation between adjacent pixels. In all cases, P ≤ 0.05 was considered significant. 

Histologic Analysis

NV incidence and severity were determined for each animal studied by functional MRI and their unexamined littermates by examination of adenosine diphosphatase (ADPase)–stained flatmounts, as previously described. ^{3 4 24 25} Severity was determined only from retinas with some degree of NV. To determine the severity of NV, three investigators independently scored each ADPase-stained retinal flatmount using the clockhour method of measuring of NV, in a masked fashion. The median number of clockhours per retina recorded by the three investigators is reported. A clock face was mentally superimposed on the retinal surface and the number of clock hours (a score from 0 to 12) occupied by abnormal vessel growth determined. To compare the severity, a two-sample Mann-Whitney rank sum test (two-sided) was used. To compare the incidence, a χ² test was performed (2 × 2). P < 0.05 was considered significant. The clockhour scoring system, which has been shown to be a valid method for quantifying retinal NV compared with counting cell nuclei above the internal limiting membrane in histologic sections, ²⁶ was used in this study. To determine the extent of the central acapillary region, the image of an ADPase-stained flatmount was captured by a charge–coupled device (CCD) camera and analyzed with NIH Image (available by ftp at zippy.nimh.nih.gov/ or at http://rsb.info.nih.gov/nih-image; developed and provided in the public domain by Wayne Rasband, National Institutes of Health, Bethesda, MD). Distances from the optic nerve head to fully vascularized peripheral retina were also measured in each clockhour for each animal. This provides a measure of the range of distances from the optic nerve head that were not fully vascularized (i.e., acapillary). Note that only a subset of retinas analyzed histologically were studied by MRI. 

Results

Model Characteristics

Representative retinal ADPase-stained flatmounts are illustrated in Figure 1 . As expected, only the P17 OIR group exhibited a capillary-free region of central retina (22% ± 3% of entire retinal area, mean ± SD). Figure 2 summarizes, from 21 pups, the range of distances from the optic nerve head to the posterior edge of the peripheral vascularized retina (i.e., the most lateral aspect of the capillary-free zone). In addition, only P17 OIR mice demonstrated significant (P < 0.05) retinal NV incidence and severity compared with age-matched controls (Table 1) . 

Magnetic Resonance Imaging

A summary of the blood parameters measured during a 2-minute carbogen challenge is present in Table 2 . Although there were some significant (P < 0.05) differences between groups, in all cases the blood gas values were within the expected range for carbogen breathing. ^{3 4 19} The significant difference (P < 0.05) in arterial pH between P17 and P34 groups appeared to be related to age and not to experimental group. These differences are not expected to confound data interpretation. As shown in Figure 2 , between OIR and age-matched control groups, retinal oxygenation response was not significantly different (P > 0.05) in the peripheral vascularized retina (i.e., >1.5 mm from the optic nerve head [ONH]). In contrast, between 0.5 and 1.5 mm from the ONH (central retina), a supernormal (P < 0.05) ΔPo ₂ was found. By P34, there was no difference (P > 0.05) in average panretinal ΔPo ₂ between control and OIR groups (Fig. 3) . 

Discussion

In the present study, the spatial pattern and temporal evolution of retinal ΔPo ₂ were determined for the first time during (P17) and after (P34) the appearance of retinal NV in a murine OIR model. Note that the small size of the newborn mice (≤3 g) effectively limited these studies to mice P17 and older. Systemic differences in arterial blood gas levels between control and experimental groups do not seem to explain the present results. For example, P_aO₂ was not significantly different between the control and experimental P17 mouse pups, but ΔPo ₂ over only the avascular retina was significantly higher than normal (Fig. 2) . Also, although P_aO₂ was different between the P34 control and experimental pups, retinal ΔPo ₂ was not different between these two groups (Fig. 3) . The present data highlight MRI and a carbogen challenge as a potentially effective noninvasive approach for studying retinal ΔPo ₂ in vivo in eyes that are too small for the application of existing techniques. 

In the control groups, retinal ΔPo ₂ at P17 was significantly lower than that at P34 (Fig. 3) . In adult rats, during a carbogen challenge, oxygen from the choroid is mostly consumed by the inner plexiform layer and does not enter the preretinal vitreous space to a substantial degree. ^{27 28} In principle, during a carbogen challenge, reduced oxygen supplied by the retinal circulation and/or increased retinal oxygen consumption could produce a subnormal response. Several interrelated hemodynamic parameters can modulate oxygen supply during a carbogen challenge, including abnormal retinal vessel autoregulation, perfusion reserve, ocular perfusion pressure, plasma pH, and vascular density. Caution is necessary when extrapolating between the rat and mouse because the choroidal contribution to the vitreal ΔPo ₂ in control and experimental mice is currently unknown. We speculate that the differences in response between P17 and P34 control groups in the present work are due to age-related changes in a combination of such hemodynamic parameters and perhaps to adjustments in oxygen consumption capacity as well. Additional studies are now needed to define the age-related changes in these factors. In disease, it seems less likely that oxygen consumption capability is increased so that measurement of subnormal retinal ΔPo ₂ could represent primarily a surrogate measure of a decrease in oxygen supply by the retinal circulation. ^{3 4 17 23} In any event, age-appropriate controls are needed for appropriate retinal ΔPo ₂ comparisons. 

In the OIR groups, different spatial response patterns were found. We assume that normal retinal ΔPo ₂ in the peripheral retina at P17 and panretinally at P34 indicates that the retinal circulation is functioning properly. To understand the regional differences in retinal ΔPo ₂ over the central and far peripheral retina at P17, two possible (not mutually exclusive) explanations were considered. First, ocular perfusion pressure could have been lower than normal, resulting in a greater than normal oxygen supply during the carbogen challenge. ^{29 30} We cannot completely rule out this possibility, because ocular perfusion pressure was not measured in this study. However, the normal peripheral retinal ΔPo ₂ at P17 in OIR mice suggests a relatively small contribution of a lower perfusion pressure to ΔPo ₂. 

A second possibility is that oxygen from the choroidal circulation entered the preretinal vitreous space producing a larger than expected response. As discussed earlier, most of the oxygen supplied to the preretinal vitreous space during the inhalation challenge is from the retinal and not the choroidal circulation. ²⁷ In the present study, if retinal oxygen consumption decreased substantially (e.g., because of a lower than normal retinal thickness or metabolically compromised retina compared with control mice), it is possible that more choroidal oxygen than normal enters the preretinal vitreous space during carbogen breathing. In the mouse OIR model at P17, the association between the central acapillary retina and a supernormal oxygenation response was consistent with an abnormal influx of choroidal oxygen. This pathophysiology appeared reversible, because by P34, after the central region has become vascularized, normal ΔPo ₂ was measured. Gu et al. ³¹ found that the thickness of all layers of the retina were preserved if mice were exposed to 75% oxygen from P7 to P27. This suggests a potentially important role for abnormally low metabolism in the central retina. More work is needed for a better understanding of the pathogenesis of this supernormal response. 

One motivation for this study was to compare the spatial and temporal patterns of retinal ΔPo ₂ in two common retinal NV models: the mouse OIR and a newborn rat model involving a variable oxygen protocol. We thought that because a common histopathology (retinal NV) was produced in these two models, a similar spatial pattern and temporal evolution of retinal ΔPo ₂ might also be observed. In contrast to the present results, in a newborn rat retinal NV model, only subnormal panretinal retinal oxygenation responses were measured during and after the appearance of retinal NV. ^{3 4} However, given the limits of the present study (discussed earlier), we could not rule out the possibility that in the mouse OIR model a subnormal response occurred earlier than P17 or was masked by choroidal oxygen, for example. Nonetheless, there are substantial differences in the type, timing, and location of injury to the retinal vessels in these models. The present study raises the possibility for the first time that the different retinal ΔPo ₂ patterns reflect an underlying difference in the type of injury that is predominant in each model. More detailed work is now needed to better our understanding of the injuries produced by the two retinal NV models. 

 Supported by National Eye Institute Grant EY10221 (BAB).

 Submitted for publication April 28, 2003; revised July 11 and August 23, 2003; accepted August 20, 2003.

 Disclosure: R. Roberts, None; W. Zhang, None; Y. Ito, None; B.A. Berkowitz, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Representative ADPase-stained retinal flatmounts from (A) P17 control mice, (B) P17 OIR mice, and (C) P34 OIR mice. Compared with the full retinal vascularization in (A) and (C), a large acapillary region was present in the central retina in (B). (B, right) Higher magnification view of the region of interest (left). Arrowheads: representative retinal NV tufts.

Figure 2.

 

Top: Histogram summary of central retina acapillary distribution in P17 OIR mice. Using ADPase-stained retinal wholemounts, the distance from the ONH to fully vascularized peripheral retina was measured in each clockhour for each animal. This provides a measure of the range of distances from the ONH that were not fully vascularized (i.e., acapillary). Note that after approximately 1.6 mm from the ONH, the far peripheral retina was vascularized in all OIR mice. Bottom: Plots of the average retinal pixel ΔPo ₂ values versus distance from the ONH for P17 control (•, n = 6) and OIR ( , ▵; n = 6) mice. In all cases, a linear relationship (P < 0.05) was found. Different symbols were used in the OIR group to highlight an apparent spatial association between retinal regions containing mixed vascular and avascular retina (0.5–1.6 mm from the ONH; see histogram) and supernormal responses ( ), as well as between the vascularized peripheral retina (1.6–2.5 mm from the ONH) and normal ΔPo ₂ (▵).

Table 1.

 

Histology Summary

Groups	NV Incidence (%)	NV Severity (Clockhours)
Control P17 Mice	0 (0/20)	0
OIR P17 Mice	100 (36/36)*	4 (1–7)*
Control P34 Mice	0 (0/8)	0
OIR P34 Mice	0 (0/9)	0

*  Significantly different from the age-matched control group; P < 0.05.

Table 2.

 

Summary of Systemic Physiology

Group	PaO2 (mm Hg)	PaCO2 (mm Hg)	pH	Core Temperature (°C)
Control P17 Mice (n = 6)	465 ± 15	58 ± 1	7.24 ± 0.01	35.8 ± 0.3
OIR P17 Mice (n = 6)	501 ± 17	59 ± 2	7.25 ± 0.01	35.1 ± 0.1
Control P34 Mice (n = 8)	527 ± 32	59 ± 1	7.20 ± 0.01* , †	35.8 ± 0.3
OIR P34 Mice (n = 7)	598 ± 3* , †	57 ± 1	7.18 ± 0.01* , †	35.9 ± 0.3

 Data are the mean ± SD.

*   P < 0.05, significantly different from P17 control value.

†   P < 0.05, significantly different from P17 OIR value.

Figure 3.

 

Summary of average retinal ΔPo ₂ over vascularized peripheral retina in control and OIR groups at P17 (left two bars) and P34 (right two bars). In addition, at P34, normal average retinal ΔPo ₂ was also measured panretinally (data not shown). No significant differences (P > 0.05) were found between control and OIR mice at either time point. The number of animals used to generate these data are listed above each bar. Error bars, SEM.

The authors thank Rod Bruan and the reviewers for thoughtful comments.