Rod Photoreceptor Function Predicts Blood Vessel Abnormality in Retinopathy of Prematurity

Abstract

purpose. To test the hypothesis that early rod dysfunction predicts the blood vessel abnormalities that are the clinical hallmark of retinopathy of prematurity (ROP).

methods. Two rat models of ROP, induced by exposure to alternating 50%/10% oxygen (50/10 model) from postnatal day (P) 0 to P14, or exposure to 75% oxygen (75 model) from P7 to P14, and controls reared in room air were studied. In a longitudinal design, electroretinographic (ERG) records and digital fundus images were obtained at P20 ± 1, P30 ± 1, and P60 ± 1. Rod sensitivity was derived from the ERG a-wave. Integrated curvature for the arterioles was calculated using Retinal Image multi-Scale Analysis (RISA) software.

results. In both ROP models, rod sensitivity was low at P20. Sensitivity improved by P60 in the 50/10 model, but remained low in the 75 model. Integrated curvature was high at P20 in both ROP models, decreased nearly to normal by P30 in the 50/10 model, but remained high in the 75 model, even at P60. At P20, rod sensitivity correlated with integrated vessel curvature. Furthermore, low rod sensitivity at P20 predicted abnormal retinal vasculature—that is, high integrated curvature—at P30 and P60. In contrast, vessel curvature at P20 did not predict sensitivity at P30 or P60.

conclusions. The rods may instigate the vascular abnormalities that are the clinical hallmark of ROP.

The ophthalmologist diagnoses retinopathy of prematurity (ROP) based on the appearance of the retinal blood vessels. For instance, dilated venules and tortuous arterioles at the posterior pole are recognized signs of high risk ROP. ¹ However, ROP is characterized not only by abnormal retinal vasculature, but also by lasting dysfunction of the neural retina, ^{2 3 4 5 6} including the rod photoreceptors which are particularly susceptible to both too much and too little oxygen. ⁷ Deficits in rod sensitivity are documented in infants and children with a history of ROP. ^{2 4 5} In rat models of ROP, abnormalities of rod structure and function antedate the retinal vascular abnormalities ⁸ and persist after the vascular abnormalities have resolved. ^{9 10}  

As in the preterm infant, ¹¹ newborn rats have incomplete vascular coverage of the retina, and the neural retina is immature. ^{12 13} Rat models of ROP are created by exposing infant rats to periods of relatively high and low oxygen during the first weeks after birth. Different schedules of oxygen exposure produce a range of effects on the retinal vasculature and on the neural retina (Akula JD et al. IOVS 2006;47:ARVO E-Abstract 3218), ^{14 15 16 17} as observed in the broad scope of human ROP. 

In rats with oxygen-induced retinopathy and significant rod dysfunction, ^{8 18} Fulton et al. ¹⁸ found that the rod outer segments were disorganized, but the total amount of rhodopsin in the retina did not differ from control rats. Dembinska et al. ^{19 20} demonstrated dysfunction of rod- and cone-driven retinal responses, as well as thinning of the outer plexiform layer. Liu et al. ^{9 10} found lasting rod photoreceptor dysfunction but resolution of the retinal vascular abnormalities. Thus, defects in the structure and function of the rods and abnormalities of the retinal blood vessels both occur in rat models of ROP. Although rod dysfunction and vascular abnormality are associated, their relationship, if any, remains to be defined. 

We studied two prominent models of ROP, along with controls, and used noninvasive techniques that enabled longitudinal assessment of the neural retina and the retinal vasculature from before weaning to adulthood. We derived receptor and postreceptor sensitivity using the electroretinogram (ERG). In the same animals in the same experimental sessions, we applied image analysis software to images of the fundus to obtain unbiased measures of the retinal vasculature. We made prospective measures at the same time points in every subject, and compared vascular and neural courses in ROP and control rats. We tested the hypothesis that early rod dysfunction predicts the blood vessel abnormalities characteristic of ROP. In addition, we evaluated postreceptor function for significant relation to vascular parameters. 

Methods

Subjects

Thirty-seven Sprague-Dawley albino rats (Charles River Laboratories, Inc., Worcester, MA) from 10 litters were studied. Litters were run in pairs. To minimize the effects of mother and birth order, on the day of birth, pups were mingled and an equal number (±1 if an odd number) returned to each dam. Pups were weaned at postnatal day (P) 25. 

All tests were performed in all subjects at P20 ± 1, P30 ± 1, and P60 ± 1. One subject died after testing at P30 and was replaced by a litter mate. The light cycle was 12 hours dark and 12 hours light (50–100 lux). All procedures were approved by the Animal Care and Use Committee at Children's Hospital Boston, and were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Procedures

Induction of Retinopathy.

Each litter, with nursing dam, was assigned to one of three oxygen-exposure paradigms. In the first (Fig. 1A) , according to the method of Penn et al., ¹⁶ on the day of birth until P14, newborn pups and dam were placed together in an oxygen-controlled environment (OxyCycler; Biospherix Ltd., Redfield, NY) where the ambient oxygen concentration was alternated every 24 hours between 50% ± 1% and 10% ± 1%. These subjects are referred to as the 50/10 model. In the second paradigm (Fig. 1B) , the ambient oxygen was maintained at a constant 75% ± 1% from P7 to 14 (the 75 model). ²³ On P14, the rats were returned to room air (21% oxygen). In preliminary studies, Akula et al. (IOVS 2006;47:ARVO E-Abstract 3219) found that the two models of ROP produced a wide range of effects on the neural retina and the retinal vasculature. In the third paradigm rats were reared in room air as controls. Each group (50/10 model, 75 model, control) contained 12 pups. 

Blood oxygen levels for the 50/10 model mimic those experienced by preterm infants. ¹⁶ However, at P0, when the 50/10 exposure began, the rat retina is quite immature, as only ganglion cells are differentiated. At P7, when the high oxygen exposure for the 75 model began, the rat retina is more mature, and photoreceptors are differentiated, as in the preterm infant. 

Electroretinography.

Before ERG testing, rats were dark adapted for a minimum of 2.5 hours. Preparations were made under dim red illumination. Subjects were anesthetized with a loading dose of 75 mg · kg⁻¹ ketamine and 7.5 mg · kg⁻¹ xylazine injected intraperitoneally. At P30 and P60, this was followed, if needed, by a booster dose (50% of loading dose) administered intramuscularly. The pupils were dilated with 2 drops of a combination of 1% phenylephrine hydrochloride and 0.2% cyclopentolate hydrochloride (Cyclomydril; Alcon, Fort Worth, TX). The corneas were anesthetized with 2 drops of 0.5% proparacaine hydrochloride. A Burian-Allen bipolar electrode (Hansen Laboratories, Coralville, IA) was placed on the left cornea. The ground electrode was placed on the tail. 

As described previously, ^{8 9 10 22 24} ERG responses to a range of brief (<1 ms) white flashes were recorded over a >5-log-unit range. The stimuli (Novatron, Dallas, TX) were delivered through a 41-cm integrating sphere at a rate that did not attenuate subsequent response amplitudes. At the upper limits of the intensity range, the interstimulus interval was >2 minutes. Stimulus intensity was controlled by calibrated neutral-density filters (Wratten filters; Eastman-Kodak, Rochester, NY). ERG records were amplified, digitized, and stored (UTAS-E 2000 system; LKC, Inc., Gaithersburg, MD). For small responses, several records were averaged. The unattenuated flash, measured at the position of the rat's eye using an integrating radiometer (S350; United Detector Technology, Orlando, FL) produced ∼40,000 μW · cm⁻² at the cornea. Since the end-on collecting area of the rat rod that is affected by ROP is unknown, stimuli are expressed in microwatts per square centimeter or normalized units, rather than photoisomerizations of rhodopsin per rod (R*). 

To evaluate photoreceptor function, the Hood and Birch ²⁵ formulation of the Lamb and Pugh ^{26 27} model of the processes involved in the activation of phototransduction was fitted to the a-waves elicited by the five brightest flashes. The a-wave results from the suppression of the circulating current of the photoreceptors. ^{28 29} Fitting was restricted to the a-wave trough or to 10 ms after the flash, whichever came first. The function takes flash intensity, i (microwatts per square centimeter), and elapsed time, t (seconds), as its inputs, so that  

\[R(i,\ t){=}{\{}1{-}\mathrm{exp}{[}{-}1/2{\cdot}i{\cdot}S{\cdot}(t{-}t_{\mathrm{d}})^{2}{]}{\}}{\cdot}Rm_{\mathrm{P}3}\ \mathrm{for}\ t\ {\geq}\ t_{\mathrm{d}},\]

where S is a sensitivity measure (μW⁻¹ · cm² · sec⁻²), t _d is a delay (seconds) of ∼3.5 ms, and Rm _P3 is the saturating amplitude of the photoreceptor response (in microvolts). S is an amplification constant based on the time constants involved in the activation of phototransduction; Rm _P3 is proportional to the number of channels in the outer segment membrane available for closure by light. 

The amplitude of the b-wave was measured from the trough of the a-wave to the peak of the b-wave. The stimulus–response function  

\[V(i)/Vm{=}i/(i{+}{\sigma})\]

was fitted to the b-wave amplitude. In this equation, V is the amplitude of the b-wave response (microvolts) to a stimulus of i intensity (microwatts per square centimeter), Vm is the saturated amplitude of the b-wave (microvolts), and σ is the flash intensity (microwatts per square centimeter) which produces a b-wave with an amplitude of half Vm. Responses to high flash intensities at which substantial a-wave intrusion occurred were not included in the fit. ³⁰ At these intensities, under dark adapted conditions, the b-wave reflects mainly the activity of the rod bipolar cell. ^{29 30 31 32 33} The log of σ was taken as representative of the sensitivity of the postreceptor ERG response. Vm is proportional to the saturating amplitude of the rod bipolar cells' light response. 

Analysis of Fundus Images.

Digital photographs of the fundus of the right eye were obtained ⁹ (RetCam; Clarity Medical Systems Inc., Dublin, CA) immediately after each ERG session. In composite images, the major arterioles and venules were identified, and three each were selected at random for analysis with Retinal Image multi-Scale Analysis (RISA) software. ^{34 35 36} Selected vessels were cropped from the main image (Fig. 2A)and segmented individually. The output was a black (background) and white (vessel) image (Fig. 2B) . In 18 of the 108 composite images, fewer than three arterioles and three venules were measurable using RISA; however, at least one arteriole and one venule were measurable in every image. The segmented image was manually edited to remove extraneous features such as the background choroidal vasculature. From each edited black and white image, RISA constructed a skeleton and marked terminal and bifurcation points (Fig. 2C) . A path was selected through the vascular “tree” by following the thicker “branch” at each bifurcation. If the thickness of two branches appeared the same, one was selected by flip of a coin. RISA calculated three geometrical properties for the selected path (Fig. 2D) : integrated curvature (IC), tortuosity index (TI), and diameter (D). IC is the sum of angles along the vessel, normalized by the vessel length (radians · pixel⁻¹). TI is the vessel length divided by the length of the straight line connecting its start and end points (dimensionless). A theoretical straight vessel has IC = 0 and TI = 1. D is the total area of the vessel divided by its length (pixels). Any departures from linear course were captured by both IC and TI; however, TI could not distinguish between the gradual, wide curves often seen in healthy blood vessels and sharply winding courses. Thus, as found in human infants, ³⁵ IC better captures the vessel appearance that a clinician would be likely to designate as “tortuous.” For each subject, mean IC, TI, and D were calculated independently for the arterioles (IC _A, TI _A, D _A) and venules (IC _V, TI _V, D _V). 

Statistical Analyses

ERG (S, Rm _P3, log σ, Vm) and RISA (IC, TI, D) parameters were evaluated by respective repeated-measures analyses of variance (ANOVA) with factors age (P20, P30, P60) and group (50/10 model, 75 model, control). For blood vessel parameters, the additional factor vessel (arteriole, venule) was included. Post hoc analyses were performed using the Tukey honestly significant difference (q) statistical test. For every subject, the rates of change for each ERG and RISA parameter were calculated using linear regression. To test rod response parameters at young ages for significant relation to vascular parameter outcomes at older ages, Pearson product moment correlations were calculated between ERG a-wave parameters at P20 and IC _A at P20, P30, and P60. The significance level (α) for all tests was P < 0.05. 

Results

Figure 3Ashows sample fits of equation 1to the ERG a-wave responses in the 50/10 model, the 75 model, and control rats obtained at P20. Figure 3Bshows sample rod-driven b-waves ³⁸ from these same animals, and Figure 3Cshows plots of the fits of equation 2to the response versus intensity relationship of those b-waves. Both ROP models demonstrate relatively diminished rod photoreceptor and postreceptor response sensitivities and amplitudes. 

The sensitivity parameters for the rod photoreceptor (S) and postreceptor retina (log σ) are plotted as a function of age in Figure 4 . At P20, rod sensitivity in 50/10 model rats was less than 75% of that in control rats, but, as in the controls, S increased with age. S in the 75 model was also low at P20, showed some recovery by P30, but unlike the 50/10 model, (or controls) made no further improvements by P60. In control rats, postreceptor response sensitivity (log σ) changed little with age. Like S, log σ at P20 showed deficits in both ROP models. In 50/10 model rats, log σ recovered rapidly, reaching nearly control values by P30; the 75 model rats recovered log σ more slowly. At P20, deficits ^{5 9 39} in postreceptor sensitivity were larger than in photoreceptor sensitivity (t = 9.13; df = 23; P < 0.001). However, with the exception of one 75 model rat, all ROP rats' log σ values fell within the range of the controls' at P60. 

ERG rod and postreceptor sensitivity and amplitude parameter values at each age, and results of analyses of variance on these data, are presented in Table 1 . The results of the ANOVA indicate that S was significantly lower in ROP rats than in controls and that, across groups, S increased significantly with age. There were significant deficits in log σ in both ROP models, and log σ significantly improved in both models by P60. Rod photoreceptor response amplitude (Rm _P3) and postreceptor response amplitude (Vm) were significantly below the controls in the ROP models, and both parameters became significantly smaller with age across all groups. There was no recovery in either ERG amplitude parameter in ROP rats. 

Figure 5shows the main steps in RISA performed on sample arterioles in the P20 50/10 and 75 model rats. In these ROP rats, both IC and TI were high compared with the values in the controls (Fig. 2) . 

IC is plotted as a function of age for arterioles and venules in Figure 6 . IC at P20 was high in ROP rats for both arterioles and venules, though the values for IC _A were much higher than for IC _V. IC _A in the 50/10 model rats recovered to control values by P30. In the 75 model rats IC _A remained high, even at P60. 

RISA parameters at each age and results of analyses of variance on these data are presented in Table 2 . IC was significantly higher in ROP rats than in controls. This effect was significantly more pronounced for IC _A than IC _V. The effects were similar for TI, though much less pronounced than for IC. In the 50/10 model, there were dramatic improvements in IC _A and a significant effect of age. D did not vary with group, but decreased significantly with age. 

As indicated in Figures 4 and 6(and Tables 1 and 2 ), in the 50/10 model rats, large improvements in the ERG postreceptor sensitivity parameter (log σ) occurred concomitantly with large improvements in the tortuosity of the retinal arterioles, IC _A. The rate of change in each individual, summarized by a linear model over the ages tested, showed that changes in both log σ and IC _A occurred most rapidly in the 50/10 model and least rapidly in the controls. The 75 model showed intermediate change in both parameters (Fig. 7) . Liu et al. ⁹ suggested that neural and vascular remodeling ⁴⁰ occur cooperatively. Indeed, across all groups, there was a significant correlation in the rates of change in log σ and IC _A (r = 0.29; P = 0.045). 

Figure 8shows S at P20 plotted against IC _A, at each test (P20, P30, P60). As shown, at P20 low rod sensitivity correlated significantly with high blood vessel abnormality, and low S at P20 also predicted vascular abnormality at P30 and P60. Of interest, the strength of the correlation increased with age. Table 3shows Pearson product moment correlations between rod parameters (S, Rm _P3) and arteriolar curvature at each test. High IC _A at P20 did not predict low S at P30 or P60. Rm _P3 did not correlate with IC _A at any age. The postreceptor sensitivity parameter (log σ) that presumably depends in part on rod sensitivity, ⁴¹ also correlated with IC _A at P20 (r = 0.57; P < 0.001). However, log σ did not predict IC _A at P30 and P60. 

Discussion

Both the neural retina and its vasculature were altered in these rat models of ROP. At age 20 days, both 50/10 and 75 model rats showed significant dysfunction in receptor and postreceptor electroretinographic responses, and both had significantly higher IC _A than did the control rats, indicating more tortuous retinal arterioles. However, rod and postreceptor sensitivity substantially improved in 50/10 model rats, as did the retinal vasculature. In contrast, the 75 model rats showed only modest improvement in rod sensitivity and arteriolar curvature, whereas postreceptor sensitivity returned to normal in all but one rat. 

We found accordant rates in the recoveries in log σ and IC _A (Fig. 7) . There are at least two not necessarily exclusive explanations for this finding. First, in the rat, the retinal vasculature supplies oxygen to the inner retina from the outer plexiform layer inward, including to the rod bipolar cells. ⁴² Assuming that the IC _A provides an indication of retinal circulation, then the recovery of log σ could be due to improvements in IC _A. Second, it is known that retinal neurons reorganize pathways extensively in instances of photoreceptor disease. ⁴⁰ Several hypoxia-induced growth factors secreted by neuroglia mediate the development of the retinal vasculature and the developing neural retina. ^{43 44 45 46 47 48} The neurovascular congruency demonstrated in the recoveries of log σ and IC _A (Fig. 7)is consistent with shared molecular patterning mechanisms ⁴⁹ such as have been demonstrated in normal retinal development. ⁵⁰ Studies of growth factor expression in ROP rats with well characterized retinal function and vasculature are a promising approach to unravelling the cooperative relationship of neural and vascular development in ROP. 

In the normally developing eye, vascular coverage becomes complete at about the same age as developing rod outer segments first appear throughout the retina, even at the periphery. ^{21 51} It is recognized that the oxygen demands of the rod photoreceptors drive the need for retinal vascularization. ⁴³ Indeed, as illustrated in Figure 9 , the improvements in S and in IC _A in the 50/10 model and the persistence of low S and high IC _A in the 75 model, went hand in hand. However, the role of the choroid in determining rod function was not addressed in this study. The choroidal vasculature supplies the photoreceptors. ⁵²  

In addition to establishing, for the first time, a consistent neurovascular relation in the ROP disease process (Fig. 9) , these data provide new evidence that the rod photoreceptors play a role in the ROP disease process. That is, that rod cell dysfunction instigates the abnormal retinal vasculature that is the clinical hallmark of the disease, and not the other way around. Specifically, rod sensitivity at P20 predicted vascular abnormality at P30 and P60 (Fig. 8 , Table 3 ). In contrast, vascular abnormality at P20 did not predict rod function at P30 and P60 (Table 3) . The temporal priority ⁵³ of rod dysfunction in these ROP rats is evidence that the photoreceptors instigate the processes which produce the abnormal retinal vasculature. Experimental manipulation of rod sensitivity is necessary to establish a causal role for rods in the pathogenesis of ROP. 

Our own past efforts to characterize the retinal vasculature in ROP rats have focused on TI, manually assessed. ⁹ Consistent with the findings of Gelman et al. ³⁵ in human infants with ROP, in the present study the IC parameter detected differences between groups (50/10 model, 75 model, controls) with greater confidence than did TI (Table 2) . However, care should be taken in the interpretation of the IC parameter, because its unit, radians per pixel, is resolution dependent. That is, the number of pixels representing the vessel in the digital image affects the calculated value of IC: The more pixels, the lower IC. To make direct comparisons of IC from camera to camera, pixels must be converted into a calibrated spatial unit, such as millimeters. We have not attempted to evaluate the real physical dimensions of the retinal vasculature in our fundus images. Nevertheless, since unit scale is irrelevant in correlation, the several significant correlations found herein between electroretinographic parameters and IC _A are robust to changes in image-acquisition method. However, the ERG is a panretinal potential, and we limited our investigation of IC _A to the posterior pole. 

High oxygen administered for the acute cardiopulmonary care of tiny, prematurely born infants has for more than a half-century been associated with ROP. ^{54 55 56 57 58 59} and one feature that is unique to the retina is the presence of the rod photoreceptors, cells that are the most demanding of oxygen in the body. ⁵² The abnormal retinal blood vessels that characterize ROP first appear within a narrow preterm age range when the developing rod outer segments are elongating rapidly and the rhodopsin content of the retina is increasing, presumably leading to burgeoning energy demands in the retina. ^{60 61} Provocatively, patients with retinitis pigmentosa are protected against oxygen-induced vasoproliferative retinopathies, as are mice with photoreceptor degenerations. ⁶² This preponderance of coincidence indicates that the metabolic demands of the rod photoreceptors play an important role in the pathogenesis of ROP. The findings of the present study further support the hypothesis that rod-driven anoxia is an “adequate and elegant” factor explaining the pathogenesis of ROP. ⁶³ If true, then treatments which “rest” the rods may be novel and efficacious therapies for ROP. 

 Supported by grants from the Knights Templar Eye Research Fund, the Fight for Sight, the March of Dimes Birth Defects Foundation, the Pearle Vision Foundation, and the Massachusetts Lions Eye Research Fund.

 Submitted for publication February 16, 2007; revised April 2 and 20, 2007; accepted June 11, 2007.

 Disclosure: J. D. Akula, None; R.M. Hansen, None; M.E. Martinez-Perez, None; A.B. Fulton, 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.

 

Experimental design. Two models of ROP were created by exposure of pups to different ambient oxygen environments during the ages when the rhodopsin content of the retina was developmentally increasing. ERGs and fundus photographs were obtained at P20 ± 1, P30 ± 1, and P60 ± 1 (gray bars). (A) The 50/10 model: starting on the day of birth, ambient oxygen concentrations were alternated every 24 hours between 50% ± 1% and 10% ± 1% for 14 days. Room air is 21% oxygen. (B) The 75 model: oxygen was maintained at 75% ± 1% from P7 to P14. (C) Growth curve for rhodopsin. ^{21 22} Rhodopsin content at P20 is ∼60% of adult.

Figure 2.

 

Sample image of the retina and steps in RISA. (A) A composite image of the retina in a P20 control rat. The posterior pole (black circle) is defined as the region within the circle bounded by the vortex veins (solid arrowheads) and concentric to the optic nerve head (open arrowhead). The meridional line ³⁷ (arrows) runs across the retina inferior to the ONH and demarcates the dorsal and ventral retina. An arteriole is selected (white box) for analysis. (B) The selected arteriole is segmented. (C) Vessel skeleton, with terminals (blue stars) and bifurcation (red stars) marked. (D) The vessel is tracked and measured. Black line: connects its start and endpoints. RISA values for IC (radians per pixel), TI, and D (pixels) for this arteriole are shown. In panels (C) and (D) the centralmost segment of arc is the boundary between two images and was not used in the analysis.

Figure 3.

 

Sample ERG records from the 50/10 model, the 75 model, and control rats obtained at P20. An artifact from the amplifier has been removed from all traces. (A) Fits of equation 1(black lines) to the ERG responses (gray lines with dots) to bright flashes producing ∼1250 to ∼20,000 μW · cm⁻². Fitting is restricted to the leading edge of the a-wave to a maximum of 10 ms after the flash. Rod sensitivity (μW⁻¹ · cm² · sec⁻²) and Rm _P3 (microvolts) are given in each panel. (B) Sample b-waves from the same animals. The numbers to the left of each row of traces indicate the stimulus intensity. Arrowheads indicate the delivery of the stimulus. Only responses used to fit equation 2are shown. (C) Fits of equation 2(lines) to the amplitudes of the b-waves in (B). For the 50/10 model, Vm = 123 μV; for the 75 model, Vm = 176 μV; for the control, Vm = 272 μV. Drop lines indicate log σ for each curve. The rightward shift of the curves in the ROP rats indicates lower sensitivity.

Figure 4.

 

Mean (±SEM) receptor and postreceptor sensitivity (S, log σ) in the 50/10 model, the 75 model, and control rats. For clarity, the control rats' data are plotted at the mean test ages and the ROP rats' data are offset ±1 day.

Table 1.

 

ERG Parameters in the 50/10 Model, the 75 Model, and Control Rats

Parameter	Age	50/10 Model	75 Model	Control	ANOVA*
Rod-photoreceptor (a-wave)
S , † (μW−1 · cm2 · sec−2)	P20	20.4 (2.1)	21.5 (1.9)	28.7 (1.8)	F group = 3.35; P = 0.047
	P30	26.2 (3.6)	26.7 (2.3)	30.0 (3.0)	F age = 3.81; P = 0.027
	P60	31.0 (3.5)	24.6 (2.6)	33.9 (4.7)	F group×age = 0.73; P = 0.576
Rm P3 (μV)	P20	−247 (22)	−269 (29)	−347 (32)	F group = 4.00; P = 0.028
	P30	−247 (20)	−237 (23)	−295 (28)	F age = 9.49; P < 0.001
	P60	−190 (21)	−197 (30)	−234 (22)	F group×age = 0.52; P = .723
Postreceptor (b-wave)
Log σ, † (log μW · cm−2)	P20	−0.166 (0.329)	−0.227 (0.336)	−0.489 (0.365)	F group = 3.57; P = 0.039
	P30	−0.393 (0.343)	−0.303 (0.326)	−0.457 (0.349)	F age = 5.69; P = 0.005
	P60	−0.441 (0.336)	−0.397 (0.316)	−0.475 (0.321)	F group×age = 2.37; P = 0.062
Vm (μV)	P20	153 (28)	209 (23)	320 (29)	F group = 8.75; P = 0.001
	P30	180 (30)	237 (24)	341 (40)	F age = 7.10; P = 0.002
	P60	147 (25)	187 (18)	253 (38)	F group×age = 0.90; P = 0.472

 Data are the mean (SEM).

*  Group factor is between subjects; age factor is within subjects. Significance defined as P < 0.05.

†  The unattenuated flash intensity is estimated to produce about 135,000 photoisomerizations of rhodopsin per rod (R*) in the adult (P60) control rat. To convert table values to ∼R*, divide S by 3.375 and add 0.528 to log σ.

Figure 5.

 

Sample images of the retina in 50/10 model and 75 model rats and steps in RISA. Features are as in Figure 2 . For each model: (A) composite images of the retina at P20; (B) segmented arterioles; (C) vessel skeletons; (D) tracking and measurement of the arterioles. Parameters of each arteriole (IC, TI, D) are as shown.

Figure 6.

 

Mean (±SEM) integrated curvature for the arterioles (IC _A) and venules (IC _V) in the 50/10 model, the 75 model, and control rats. For clarity, the control rats' data are plotted at the mean test ages and ROP rats' data are offset ±1 day.

Table 2.

 

Morphologic (RISA) Parameters in the 50/10 Model, the 75 Model, and Control Rats

Parameter	Age	50/10 Model	75 Model	Control	ANOVA*
Integrated curvature (radians · pixel−1)
IC A	P20	0.0250 (0.0020)	0.0277 (0.0024)	0.0149 (0.0007)	F group = 20.63; P < 0.001
	P30	0.0169 (0.0011)	0.0239 (0.0018)	0.0176 (0.0005)	F age = 25.43; P < 0.001
	P60	0.0174 (0.0007)	0.0237 (0.0015)	0.0154 (0.0005)	F vessel = 15.57; P < 0.001
					F group×age = 6.96; P < 0.001
IC V	P20	0.0222 (0.0010)	0.0189 (0.0008)	0.0179 (0.0008)	F group×vessel = 14.23; P < 0.001
	P30	0.0181 (0.0008)	0.0173 (0.0008)	0.0168 (0.0007)	F age×vessel = 0.38; P = 0.686
	P60	0.0174 (0.0009)	0.0163 (0.0005)	0.0140 (0.0004)	F group×age×vessel = 2.62; P = 0.043
Tortuosity index
TI A	P20	1.141 (0.015)	1.128 (0.011)	1.117 (0.013)	F group = 3.32; P = 0.048
	P30	1.118 (0.009)	1.131 (0.013)	1.121 (0.011)	F age = 0.74; P = 0.480
	P60	1.134 (0.013)	1.130 (0.010)	1.108 (0.004)	F vessel = 8.85; P = 0.005
					F group×age = 0.65; P = 0.629
TI V	P20	1.128 (0.014)	1.109 (0.011)	1.100 (0.006)	F group×vessel = 0.19; P = 0.831
	P30	1.114 (0.014)	1.103 (0.009)	1.099 (0.006)	F age×vessel = 0.03; P = 0.973
	P60	1.109 (0.010)	1.109 (0.013)	1.097 (0.006)	F group×age×vessel = 0.35; P = 0.841
Diameter (pixels)
D A	P20	4.44 (0.13)	4.55 (0.29)	4.23 (0.17)	F group = 0.08; P = 0.925
	P30	3.95 (0.13)	4.17 (0.13)	4.34 (0.19)	F age = 20.78; P < 0.001
	P60	4.07 (0.09)	4.07 (0.12)	4.05 (0.09)	F vessel = 186.21; P < 0.001
					F group×age = 3.78; P = 0.008
D V	P20	5.45 (0.11)	5.37 (0.13)	4.90 (0.16)	F group×vessel = 0.65; P = 0.529
	P30	4.85 (0.14)	4.82 (0.15)	5.10 (0.09)	F age×vessel = 4.28; P = 0.018
	P60	4.46 (0.07)	4.48 (0.09)	4.62 (0.11)	F group×age×vessel = 0.66; P = 0.621

 Data are the mean (SEM).

*  Group factor is between subjects; age and vessel factors are within subjects. Significance defined as P < 0.05.

Figure 7.

 

Mean (±SEM) rates of change in log σ and IC _A for the 50/10 model, the 75 model, and control rats.

Figure 8.

 

Mean IC _A and S in 50/10 model, 75 model, and control rats expressed as proportion of the mean in controls (stippled lines). Solid lines are linear regressions through all the data; Pearson's product moment correlation (r) and significance (P) are given in each panel. In each panel, the abscissa is the sensitivity of rod responses obtained at P20, and the ordinate is IC _A at each test. Rod sensitivity at P20 predicted vascular abnormality at every age. Strength of correlation increased with age.

Table 3.

 

Pearson Product Moment Correlations of Photoreceptor and Vascular Parameters

Age	ERG Parameter	RISA Parameter: IC A , † (radians · pixel−1)
				P20	P30	P60
P20	S (μW−1 · cm2 · sec−2)	−0.300* (0.038)	−0.318* (0.029)	−0.332* (0.024)
	Rm P3 , † (μV)	0.256 (0.066)	0.076 (0.330)	0.143 (0.203)
P30	S (μW−1 · cm2 · sec−2)	−0.089 (0.303)	−0.009 (0.478)	−0.121 (0.242)
	Rm P3 , † (μV)	0.233 (0.085)	0.090 (0.301)	0.206 (0.114)
P60	S (μW−1 · cm2 · sec−2)	−0.103 (0.275)	−0.168 (0.164)	−0.217 (0.102)
	Rm P3 , † (μV)	0.208 (0.111)	−0.067 (0.349)	0.030 (0.431)

 Data are r (sig).

*  Significant correlation (P < 0.05).

†  Lower values are better.

Figure 9.

 

Mean S and IC _A replotted from Figures 4 and 6 , respectively, as proportion of the mean values in P60 controls (circled, offset for clarity). The neural and vascular patterns of recovery from ROP in the 50/10 model and 75 model rats are similar.