Narrow-Band, Long-Wavelength Lighting Promotes Hyperopia and Retards Vision-Induced Myopia in Infant Rhesus Monkeys

Abstract

The purpose of this investigation was to determine the effects of narrow band, long-wavelength lighting on normal refractive development and the phenomena of lens compensation and form-deprivation myopia (FDM) in infant rhesus monkeys. Starting at 24 and continuing until 151 days of age, 27 infant rhesus monkeys were reared under long-wavelength LED lighting (630 nm; illuminance = 274 ± 64 lux) with unrestricted vision (Red Light (RL) controls, n = 7) or a +3 D (+3D-RL, n = 7), −3 D (−3D-RL, n = 6) or diffuser lens (From Deprivation (FD-RL), n = 7) in front of one eye and a plano lens in front of the fellow eye. Refractive development, corneal power, and vitreous chamber depth were measured by retinoscopy, keratometry, and ultrasonography, respectively. Comparison data were obtained from normal monkeys (Normal Light (NL) controls, n = 39) and lens- (+3D-NL, n = 9; −3D-NL, n = 18) and diffuser-reared controls (FD-NL, n = 16) housed under white fluorescent lighting. At the end of the treatment period, median refractive errors for both eyes of all RL groups were significantly more hyperopic than that for NL groups (P = 0.0001 to 0.016). In contrast to fluorescent lighting, red ambient lighting greatly reduced the likelihood that infant monkeys would develop either FDM or compensating myopia in response to imposed hyperopic defocus. However, as in the +3D-NL monkeys, the treated eyes of the +3D-RL monkeys exhibited relative hyperopic shifts resulting in significant anisometropias that compensated for the monocular lens- imposed defocus (P = 0.001). The red-light-induced alterations in refractive development were associated with reduced vitreous chamber elongation and increases in choroidal thickness. The results suggest that chromatic cues play a role in vision- dependent emmetropization in primates. Narrow-band, long-wavelength lighting prevents the axial elongation typically produced by either form deprivation or hyperopic defocus, possibly by creating direction signals normally associated with myopic defocus.

1. Introduction:

Evidence from many species indicates that refractive development is actively regulated by visual feedback derived from the eye’s effective refractive state (i.e., optical defocus) (Graham & Judge, 1999, Howlett & McFadden, 2009, Hung, Crawford & Smith III, 1995, Schaeffel, Glasser & Howland, 1988, Shaikh, Siegwart & Norton, 1999, Siegwart & Norton, 1999, Smith III & Hung, 1999). Recent studies suggest, however, that the spectral composition of ambient lighting can potentially influence the operation of the defocus-driven emmetropization cascade in a variety of ways. First, because of longitudinal chromatic aberration (LCA), the eye is more hyperopic/less myopic for long- versus short-wavelength light. As a consequence, alterations in the wavelength composition of ambient lighting could alter the target or set point for emmetropization, especially if the goal of emmetropization is to maximize luminance contrast (Schaeffel & Howland, 1991, Seidemann & Schaeffel, 2002, Wildsoet, Howland, Falconer & Dick, 1993). The spectral composition of ambient lighting can also affect the ocular component changes that underlie the refractive errors produced by optical defocus (Rucker & Wallman, 2008). For example, in chickens reared under short-wavelength lighting that preferentially stimulates the short-wavelength and ultraviolet-sensitive cones, lens compensation is associated with changes in overall eye length without accompanying changes in choroidal thickness (Rucker & Wallman, 2008). On the other hand, in chickens reared under long-wavelength lighting that selectively stimulates the long-wavelength and double cones, lens compensation is mediated by changes in choroidal thickness with little change in overall eye length. These results suggest that these two compensating ocular responses are dominated by different cones types and, thus, dependent on the spectral composition of the ambient lighting. In addition, experiments in chickens have demonstrated that the emmetropization process can use chromatic cues from LCA to encode the sign of defocus and to regulate appropriate compensating eye growth (Rucker, 2013, Rucker & Osorio, 2008, Rucker & Wallman, 2008, Rucker & Wallman, 2009, Rucker & Wallman, 2012). Thus, it would be expected that the spectral composition of ambient lighting could influence the efficacy of vision-dependent emmetropization.

There is increasing evidence that the spectral composition of ambient lighting can influence normal ocular growth and refractive development. However, the available data are contradictory. For instance, fish (Kroger & Wagner, 1996), chickens (Rucker & Osorio, 2008, Seidemann & Schaeffel, 2002), and guinea pigs (Jiang, Zhang, Schaeffel, Xiong, Zheng, Zhou, Lu & Qu, 2014, Long, Chen & Chu, 2009, Qian, Liu, Dai, Chen, Zhou & Chu, 2013) exposed to short-wavelength lighting exhibited shorter axial lengths and/or became hyperopic in comparison to animals reared under long-wavelength light. In the earlier studies that involved relatively short treatment durations, the differences in refractive error were similar in magnitude to the wavelength-dependent differences in the eye’s focal planes produced by LCA, suggesting that the spectral composition of ambient lighting predictably changed the target for emmetropization (Kroger & Wagner, 1996, Seidemann & Schaeffel, 2002). However, when longer treatment periods were employed, the magnitude of the refractive changes observed in chickens and guinea pigs continued to increase well beyond predictions based on LCA (Foulds, Barathi & Luu, 2013, Liu, Qian, He, Hu, Zhou & Dai, 2011). Interestingly, refractive-error changes opposite in direction from predictions based on LCA have been observed in both monkeys and tree shews. Specifically, monkeys reared with long-wavelength-pass filters (Smith III, Hung, Arumugam, Holden, Neitz & Neitz, 2015) and tree shrews reared in narrow-band, long-wavelength lighting (Gawne, Siegwart Jr, Ward & Norton, 2017a) exhibited reduced axial elongation rates and increased levels of hyperopia. While the direction of refractive changes differed between some studies, the progressive nature of these ametropias indicates that the absence of chromatic cues and/or the presence of inappropriate or erroneous chromatic sign-of-defocus cues can interfere with emmetropization.

In some studies, however, substantial alterations in the spectral composition of ambient lighting have failed to alter vision-dependent changes in refractive error. In particular, it has been shown that rearing chickens in quasi-monochromatic light does not necessarily alter refractive compensation to either imposed hyperopic or myopic defocus (Rohrer, Schaeffel & Zrenner, 1992, Seidemann & Schaeffel, 2002) or the recovery from form-deprivation myopia (FDM) (Wildsoet et al., 1993). These observations indicate that luminance contrast cues can guide refractive development and that chromatic signals are not always essential, calling into question the relative significance of color signals to emmetropization and the potential impact of the spectral composition of ambient lighting.

It is important to determine how the wavelength composition of ambient lighting influences emmetropization because manipulating the spectral composition of ambient lighting may have therapeutic potential for controlling ocular growth and refractive development in humans. In this respect, it has been hypothesized that outdoor scenes are protective against the onset of myopia in children because in contrast to indoor scenes they are often dominated by short-wavelength light (Foulds et al., 2013). The rhesus monkey is the subject of choice for the proposed studies because the resulting data should be directly applicable to humans. Previous studies have shown a close correspondence, both qualitative and quantitative, between humans and macaques in the course of emmetropization (Bradley, Fernandes, Lynn, Tigges & Boothe, 1999, Kiely, Crewther, Nathan, Brennan, Efron & Madigan, 1987, Qiao-Grider, Hung, Kee, Ramamirtham & Smith III, 2007), the changes in ocular components that occur during normal development (Bradley et al., 1999, Fernandes, Bradley, Tigges, Tigges & Herndon, 2003, Qiao-Grider et al., 2007), the alterations in ocular components that are associated with ametropias,(Qiao-Grider, Hung, Kee, Ramamirtham & Smith III, 2010, Smith III & Hung, 1999, Tigges, Tigges, Fernandes, Eggers & Gammon, 1990) and the effects of vision on refractive development (Phillips, 2005, Rabin, Van Sluyters & Malach, 1981). However, to date little is known about how the spectral composition of ambient lighting affects emmetropization in non-human primates and the available data are conflicting. For example, whereas Liu et al (Liu., Hu, He, Zhou, Dai, Qu & Liu, 2014) found that emmetropization in rhesus monkeys reared in quasi-monochromatic light was relatively unaffected, studies in our lab showed that rearing rhesus monkeys with long- wavelength pass filters (i.e., red filters) consistently produced axial hyperopia (Smith III et al., 2015). The aim of this investigation was to determine if narrow-band, long-wavelength lighting interfered with normal refractive development, the henomenon of form-deprivation myopia (FDM) and/or the compensating ocular growth normally produced by imposed defocus in infant rhesus monkeys.

2. Methods

2.1. Subjects and Rearing Procedures

Rhesus monkeys (Macaca mulatta) were selected as subjects because their spectral sensitivities and tri-chromatic color vision are similar to those of humans (Bowmaker, Dartnall, Lythgoe & Mollon, 1978, DeValois, Morgan, Polson, Mead & Hull, 1974, Harwerth & Smith III, 1985, Kalloniatis & Harwerth, 1990). This is important because primate color vision evolved in an unique way compared to most eutherian mammals and very differently from that of birds (Jacobs, 2009). In particular, primates are the only eutherian mammal that has evolved a third cone photopigment and, possibly more importantly, a midget cell system in the retina that supports an antagonistic combination of inputs from these unique M- and L-cones (Jacobs, 2009). Therefore, the emmetropization process in primates may have developed ways to use chromatic cues that did not occur in other species, which might explain why alterations in the spectral composition of ambient lighting produced qualitatively different results in monkeys (Smith III et al., 2015) versus guinea pigs (Liu et al., 2011) or chickens (Foulds et al., 2013).

The primary subjects were 27 infant rhesus monkeys that were obtained at 2–3 weeks of age and initially housed in our non-human primate nursery that was illuminated by “white” fluorescent lighting (Philips tl735, CCT = 3500K) that produced an average illuminance of 480 lux (range = 342–688 lux; for husbandry details see Smith & Hung, 1999). At 24 ± 3 days of age the monkeys were transferred to a separate nursery room (3 m × 4.6 m with 12 individual cages and a large group socialization area) outfitted with a ceiling mounted light emitting diode (LED) lighting system that was rated for outdoor use and waterproof (Philips ColorGraze MX4 Powercore lighting system, Philips North America, Andover, MA). The system included 14 four-foot-long fixtures that each held 48 LEDs (12 each of the “white”, “red”, “blue” and “green” EDs). A computer algorithm was used to independently control the intensity of each color LED. For these experiments, the room was illuminated steadily and exclusively via the red LED (630 nm; 20 nm half-max bandwidth). The average illuminance, expressed in human lux, was 274 ± 64 lux (average irradiance = 1.39 W/m²) and varied from 203 to 682 lux within the lower and upper cages, respectively. As in our standard nursery room, animals were maintained on a 12 hr light/12 hr dark cycle with the lights-on cycle beginning at 7:00 am. Figure 1 shows the emission spectrum for the red LEDs and the spectral sensitivities for the macaque short-, middle-, and long-wavelength- cones (Baylor, Nunn & Schnapf, 1987) and for the melanopsin photopigment contained in intrinsically photosensitive retinal ganglion cells (Govardovskii, Fyhrquist, Reuter, Kuzmin & Donner, 2000). The red ambient lighting primarily stimulated the long- and middle-wavelength cone mechanisms. There was little or no overlap between the red LED emission spectrum and the spectral sensitivities of the short-wavelength-sensitive cones or the melanopsin photopigment contained in intrinsically photosensitive retinal ganglion cells.

Figure 1.

Comparison of the spectral characteristics for macaque cones and melanopsin with the emission spectrum for the red LEDs (peak = 630 nm; 20 nm half-maximum bandpass). The spectral sensitivities of the short-, middle-, and long-wavelength sensitive cones are represented by the dashed blue, dashed green and solid red lines, respectively. The cone spectral sensitivities are from Baylor, Nunn, and Schnapf (1987) The dashed black line shows the melanopsin absorption spectrum that was calculated using the Govardovskii et al. (2000) template for a pigment with a peak absorption at 480 nm. The spectral energy distribution of the long-wavelength ambient lighting measured in the nursery room is shown by the solid black line.

Simultaneous with the onset of red ambient lighting, the experimental monkeys were randomized into 1 of 4 treatment groups. Seven monkeys were reared with unrestricted vision (red light (RL) control group). Monocular form deprivation was induced in 7 monkeys by placing a diffuser spectacle lens (Bangerter LP “Light Perception” occlusion foil; Fresnel Prism and Lens Co, Prairie, MN) over one eye (FD-RL group). Viewing through the diffusers reduced the spatial vision of human observers to spatial frequencies below about 0.5 cycles per degree significantly reducing any potential luminance cues for determining the sign of optical defocus (Smith III & Hung, 2000). Relative monocular hyperopic (n = 6) and myopic defocus (n = 7) were induced by securing a −3 D (−3D-RL group) or a +3 D (+3D-RL) spectacle lens in front of one of eye, respectively. The diffusers and powered lenses were held in place using a light-weight helmet that has been previously described in detail (Hung, Arumugam, Ostrin, Patel, Trier, Jong & Smith III, 2018, Smith III & Hung, 1999). The helmets also secured zero-powered (plano) spectacle lenses in front of the fellow control eyes of the diffuser-and lens-reared monkeys. These lens-rearing regimens were continued in an uninterrupted manner until 151 ± 3 days of age. The treatment period encompassed the rapid phase of emmetropization in rhesus monkeys (Qiao-Grider et al., 2007), and at typical “white” fluorescent lighting levels, was long enough to promote substantial amounts of FDM (Smith III, Hung & Huang, 2012) and complete compensating growth for the degrees of imposed defocus (Hung et al., 1995, Smith III & Hung, 1999).

Comparison data, some of which has been reported in previous publications, were obtained from infant monkeys reared with monocular diffusers (n = 16; FD normal light (NL) group), monocular −3 D (n = 18; −3D-NL group) or monocular +3 D treatment lenses (n = 9; +3D-NL group) under white fluorescent lighting (Hung et al., 2018, Hung et al., 1995, Smith III & Hung, 1999, Smith III, Hung, Arumugam & Huang, 2013, Smith III, Hung, Arumugam, Wensveen, Chino & Harwerth, 2017, Smith III et al., 2012, Smith III, Hung, Huang, Blasdel, Humbird & Bockhorst, 2010). The goggles, diffusers, and powered lenses for the treated eyes, the control lenses for the fellow eyes, and the onset and duration of lens wear for lens-reared NL monkeys were identical to those for the RL monkeys. Control data were also available for 39 infant monkeys that were reared with unrestricted vision under white fluorescent lighting (NL control group) (Hung et al., 2018, Hung, Ramamirtham, Huang, Qiao-Grider & Smith III, 2008, Qiao-Grider et al., 2007, Smith III & Hung, 1999, Smith III et al., 2013). The rearing procedures, general husbandry protocols, and biometric measurement methods for the NL control animals were identical to those for the RL monkeys.

2.2. Ocular Biometric Measures

The procedures employed for measuring refractive status, corneal power, and axial dimensions have been described in detail previously (Hung et al., 2018, Smith III & Hung, 1999). Briefly, the monkeys were anesthetized (intramuscular injection: ketamine hydrochloride, 15–20 mg/kg, and acepromazine maleate, 0.15–0.2 mg/kg; topical: 0.5% tetracaine hydrochloride) and cycloplegia was induced by the topical instillation of 1% tropicamide 25 and 20 minutes prior to obtaining the measurements. The refractive state of each eye was measured independently by two experienced investigators using a streak retinoscope and averaged using matrix notation (Harris, 1988). Refractive error was defined as the spherical-equivalent, spectacle-plane refractive correction (95% limits of agreement = ±0.60 D) (Kee, Hung, Qiao-Grider, Roorda & Smith III, 2004). The anterior radius of curvature of the cornea was measured using a hand-held keratometer (Alcon Auto-keratometer, Alcon Inc., St. Louis, MO). Three readings were taken and the results for the two principal meridians were averaged. The spherical-equivalent central corneal power was calculated using an assumed refractive index of 1.3375 (95% limits of agreement = +0.49 to −0.37 D for mean corneal power) (Kee, Hung, Qiao, Habib & Smith III, 2002). When the corneal power exceeded the measurement range of the keratometer (>62 D), a corneal video topographer was used (EyeSys 2000, EyeSys vision Inc., Houston, Tx). Ocular axial dimensions were measured by A-scan ultrasonography using a 13-MHZ transducer (Image 2000; Mentor, Norwell, MA); 10 separate measurements were averaged (95% limits of agreement for vitreous chamber depth = ±0.05mm) (Smith III et al., 2012). The initial biometric measures were obtained at ages corresponding to the start of lens wear and every two weeks throughout the observation period.

For all of the RL monkeys and 8 control eyes (the right eyes of 5 NL controls and the fellow control eyes for 3 lens-reared NL monkeys), subfoveal choroidal thickness was measured with spectral domain optical coherence tomography (SD-OCT, Spectralis, Heidelberg, Germany). See Hung et al. (Hung et al., 2018) for details. All measurements were conducted between 9:00 and 11:00 am to minimize potential confounding effects of diurnal variations in choroidal thickness. The OCT scan pattern was centered and focused on the fovea and the instrument’s enhanced depth imaging mode was used to improve the visibility of the choroidal-scleral border. The scan data were exported and analyzed using custom Matlab software, which further enhanced the visibility of the choroidal-scleral interface (Girard, Strouthidis, Ethier & Mari, 2011, Patel, McAllister, Pardon & Harwerth, 2018). An experienced, masked observer manually segmented each scan to identify Bruch’s membrane and the choroidal-scleral interface. The center of the fovea was identified as the deepest point in the foveal pit. Our primary measure was sub-foveal choroidal thickness, defined as the average thickness across 1.8 to 2.1 mm segments centered at the foveal pit, after correction for lateral magnification. The 95% limits-of-agreement for between session comparisons in control eyes was −10.4 to +5.9 µm (Hung et al., 2018). The initial choroidal thickness measures were obtained at ages corresponding to the onset of lens wear; subsequent measures were obtained at several time points during the treatment period. The instrument’s auto re-scan feature was employed to ensure that all subsequent scans were performed at the same retinal location as the baseline measurements.

All rearing and experimental procedures were reviewed and approved by the University of Houston’s Institutional Animal Care and Use Committee and were in compliance with the ARVO Animal Statement and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

2.3. Statistical Methods

Statistical analyses were performed using Minitab (Release 16.2.4, Minitab Inc, State College, PA) and Super ANOVA software (Abacus Concepts, Inc, Berkeley, CA). At ages corresponding to the end of the treatment period, the distribution of refractive errors in normal monkeys is leptokurtic. Consequently, nonparametric Mann-Whitney tests were used to compare the median end-of-treatment refractive errors between pairs of subject groups. Mixed-design ANOVAs with Greenhouse-Geisser (G-G) corrections for repeated measures, followed by unpaired T tests, were used to compare changes in refractive error, anisometropia, and choroidal thickness between groups that occurred during the lens-rearing period. Paired-student T-tests were employed to examine interocular differences in ocular parameters within a given subject group. One-way ANOVAs were used to examine between-group differences in ocular parameters at ages corresponding to the start and end of lens wear. With one exception, standard deviations were used to quantify the dispersion of data; specifically, standard errors of the mean were employed for the choroidal thickness measures represented in Figure 10. The relationship between refractive error and choroidal thickness and the relationship between refractive error and the ratio between vitreous chamber depth and corneal radius were characterized using linear regression analyses.

Figure 10.

The average (±SEM) relative changes in choroidal thickness for the treated or right eyes (red symbols) and the fellow control eyes (open symbols) of red-light-reared controls (A, diamonds) and red-light monkeys reared with monocular form deprivation (B, up triangles) and −3 D (C, circles) and +3 D of imposed anisometropia (D, hexagons). The solid gray squares represent data for the untreated control eyes. E. Changes in ametropia that took place during the treatment period plotted as function of the changes in choroidal thickness for individual animals. The diamonds, up triangles, circles, and hexagons represent, respectively, the red-light controls, the red-light FD monkeys and the red-light monkeys reared with −3D and +3 D of imposed anisometropia. The treated and fellow eyes of the lens-reared treated monkeys are represented by the filled and open symbols, respectively. The solid line was determined by linear regression analysis.

3.0. Results

3.1. General Observations

The RL monkeys were observed multiple times each day, most commonly during the normal bottle feeding sessions and when the helmets and lenses were inspected (typically every two hours during the daily lights-on cycle). We did not observe any behavioral differences between the RL and NL monkeys; there were no obvious differences in their general activity levels during the daily light cycle, including during the group socialization periods. There were also no differences in weight gains between the RL and NL monkeys.

All of the RL monkeys exhibited increases in pupil size when they were housed under the red ambient lighting. Measurements of pupil diameter were obtained from infra-red video recordings at several time points during the red-light exposure period. During the first 4–5 weeks of red-light exposure (<67 days of age), the average pupil diameter for the control eyes of RL monkeys (n = 27) was significantly larger than that for age-matched NL monkeys (n = 7) housed under white fluorescent lighting (4.68 ± 0.46 mm versus 3.58 ± 0.43 mm, T = 5.96; P < 0.001). There was no overlap in the ranges of pupil diameters for the RL and NL monkeys. Over the observation period, the pupil diameters of the RL infants increased so that after 240 days of age, the average pupil size had increased by 0.31 mm. Similarly, the pupil diameters of the NL animals increased with age; however, the pupils for the older RL animals were still significantly larger than those for age-matched NL controls (4.99 ± 0.56 mm versus 3.92 ± 0.32 mm, T = 5.44; P < 0.001). The pupils of the RL animals were reactive to changes in the level of red ambient lighting and pupil miosis could be observed when the animals’ attention was directed at near objects.

3.2. Effects of Red Light on Refractive Development

At the onset of the treatment period, there were no significant interocular differences in refractive error (F= 2.81; P = 0.06), corneal curvature (F = 1.18; P = 0.34), anterior chamber depth (F= 1.60; P = 0.22), lens thickness (F = 0.80; P = 0.51) or vitreous chamber depth (F= 0.10; P = 0.96) in any of the RL groups. Moreover, the optical and axial characteristics of the eyes of the experimental animals in all RL groups were similar to those of age-matched NL control monkeys. Most critically, there were no between group differences in refractive error (F = 0.64; P = 0.72) or vitreous chamber depth (F = 1.59; P = 0.24).

End-of-treatment refractive errors and biometric data for all groups are shown in Table 1. Rearing infant monkeys with unrestricted vision under the red ambient lighting (RL controls) resulted in relative hyperopic refractive errors. Figure 2, which shows spherical-equivalent ametropias plotted as a function of age for the right and left eyes of individual animals, compares refractive development between the RL control and NL control groups (thin grey lines). At baseline, the refractive errors for the RL control monkeys were on average slightly less hyperopic than normal, but clearly within the range of refractive errors for the NL control monkeys. Over the next several months, while the majority of NL control monkeys showed slow reductions in hyperopia, the RL control monkeys exhibited little or no change in refractive error (Fig 2A, B, D) or slow increases in hyperopia (Fig 2C, E, F,G). At the end of the observation period, the refractive errors for two of the RL control monkeys were similar to those commonly observed in NL control monkeys (Fig 2A, B). However, the refractive errors for the other 5 RL controls were more hyperopic than the majority of NL monkeys and the median refractive error for the RL control monkeys was significantly more hyperopic than that for the NL control monkeys (right eyes: +3.56 D vs +2.44 D; P = 0.008).

Table 1.

End of treatment refractive errors and biometric measures for control and experimental monkeys. NL = normal light; RL = red light.

Group	Refractive Error (D) (median)	Refractive Error (D) (mean ± SD)	Corneal Power (D) (mean ± SD)	Anterior Chamber Depth (mm) (mean ± SD)	Lens Thickness (mm) (mean ± SD)	Vitreous Chamber Depth (mm) (mean ± SD)
NL Controls Right eye Left eye N = 39	+2.38 +2.50	+2.46 ± 1.04 +2.48 ± 1.01	55.84 ± 1.59 55.93 ± 1.65	3.07 ± 0.29 3.10 ± 0.29	3.62 ± 0.21 3.60 ± 0.20	9.84 ± 0.31 9.85 ± 0.31
RL Controls Right eye Left eye N = 7	+3.56# +3.75#	+3.32 ± 0.69# +3.77 ± 1.37#	55.87 ± 2.09 55.86 ± 2.11	3.17± 0.12 3.18 ± 0.11	3.65 ± 0.07 3.62 ± 0.07	9.74 ± 0.41 9.64 ± 0.34
+3D-NL Treated eye Fellow eye N = 9	* +5.00# +3.06	* +5.02 ± 0.98# +3.13 ± 0.99	55.80 ± 1.53 56.13 ± 1.14	3.13 ± 0.15 3.03 ± 0.09	3.54 ± 0.13 3.56 ± 0.14	* 9.51 ± 0.31# 9.83 ± 0.31
+3D-RL Treated eye Fellow eye N = 7	+6.00# +4.69#	* +6.21 ± 1.91# +5.05 ± 2.18#	54.70 ± 1.39 55.04 ± 1.32	3.02 ± 0.12 3.04 ± 0.09	3.70 ± 0.11 3.70 ± 0.14	9.43 ± 0.55 9.59 ± 0.66
-3D-NL controls Treated eye Fellow eye N = 18	* −0.06# +2.38	* +0.86 ± 1.91# +3.01 ± 1.47	55.13 ± 1.49 55.02 ± 1.58	3.11 ± 0.15 3.10 ± 0.13	3.64 ± 0.12 3.65 ± 0.13	* 10.40 ± 0.55# 10.01 ± 0.49
−3D-RL Treated eye Fellow eye N = 6	+5.97# +6.25#	+5.31 ± 2.53# +5.38 ± 1.89#	56.08 ± 0.99 55.96 ± 1.16	3.01 ± 0.09 2.99 ± 0.13	3.71 ± 0.11 3.71 ± 0.14	9.25 ± 0.63 9.32 ± 0.60
FD-NL Treated eye Fellow eye N = 16	* −1.19# +2.63	* −1.30 ± 4.28# +3.18 ± 1.68	55.98 ± 1.52 55.45 ± 1.57	3.12 ± 0.28 3.14 ± 0.24	3.52 ± 0.27 3.55 ± 0.25	* 10.60 ± 0.92 9.89 ± 0.58
FD-RL Treated eye Fellow eye N = 7	+5.00# +4.56#	+3.99 ± 4.84 +4.46 ± 1.79	54.19 ± 1.32 54.35 ± 1.46	3.03 ± 0.05 2.97 ± 0.13	3.74 ± 0.12 3.78 ± 0.17	10.08 ± 1.25 10.01 ± 0.64

^*indicates significant interocular difference (p<0.05)

^#indicates significantly different from normal monkeys (p<0.05)

Figure 2.

Spherical-equivalent, spectacle-plane refractive corrections plotted as a function of age for the right (red symbols) and left eyes (open symbols) of red-light-reared control animals. All of the animals were reared with unrestricted vision. The thin gray lines in each plot represent data for the right eyes of the 39 normal control monkeys. The panels are arranged from A to G according to the degree of hyperopia at the end of the observation period.

The red ambient lighting greatly reduced the likelihood that infant monkeys would develop either FDM or compensating myopic changes in response to imposed hyperopic defocus. Figures 3 and 4 illustrate the longitudinal changes in refractive error for representative NL and RL monkeys that were reared, respectively, with monocular form deprivation or with −3D of imposed anisometropia. The FD-NL and −3D-NL monkeys (top rows) consistently developed relative myopic refractive errors in their treated eyes. Only 1 of 16 FD-NL monkeys (Fig 3A) and 1 of 18 −3D-NL monkeys (Fig 4A) failed to exhibit relative myopic alterations in their treated eyes. In contrast, only 1 of 7 FD-RL monkeys and 1 of 6 −3D-RL monkeys showed any signs of myopic alterations in their treated eye (Fig 3N and 4L). The other FD-RL and −3D-RL monkeys either became more hyperopic in both eyes, but remained isometropic (Figure 3I-M and Figure 4H-J) or showed hyperopic shifts in both eyes, with larger changes taking place in their treated eyes (Fig 3H and Fig 4G, K).

Figure 3.

Spherical-equivalent, spectacle-plane refractive corrections plotted as a function of age for the treated (filled symbols) and fellow eyes (open symbols) of representative controls (top row) and red-light-reared monkeys (bottom row). All of the animals were reared with diffuser lenses in front of their treated eyes and plano lenses in front of their fellow eyes. The thin gray lines in each plot represent data for the right eyes of the 39 normal control monkeys. In the top row the panels are arranged from left to right according to the maximum degree of myopic anisometropia; the monkeys represented in panels A and G exhibited the smallest and largest degrees of myopic anisometropia, respectively; the monkey represented in panel D exhibited the median degree of myopic anisometropia in the control group. In the bottom row, the panels are arranged from left to right in order of decreasing hyperopia in the lens-reared eye.

Figure 4.

Spherical-equivalent, spectacle-plane refractive corrections plotted as a function of age for the treated (filled symbols) and fellow eyes (open symbols) of representative lens-reared controls (top row) and red-light-reared monkeys (bottom row). All of the animals were reared with −3.0 D lenses in front of their treated eyes and plano lenses in front of their fellow eyes. The thin gray lines in each plot represent data for the right eyes of the 39 normal control monkeys. In the top row the panels are arranged from left to right according to the maximum degree of myopic anisometropia; the monkeys represented in panels A and F exhibited the smallest and largest degrees of myopic anisometropia, respectively; the monkey represented in panel C exhibited the median degree of myopic anisometropia in the control group. In the bottom row, the panels are arranged from left to right in order of decreasing hyperopia in the lens-reared eye.

In contrast, the red ambient lighting augmented compensating hyperopic shifts produced by imposed myopic defocus. Longitudinal refractive-error data are shown for representative +3D-NL monkeys (top row) and +3D-RL monkeys (bottom row) in Figure 5. The treated eyes of all 9 +3D-NL monkeys and all 7 +3D-RL monkeys became more hyperopic than their respective fellow control eyes. As observed in the other RL treatment groups, the fellow eyes of many of the +3D-RL monkeys also developed higher than normal degrees of hyperopia (Fig 5J-N). It is noteworthy that in these animals, the treated eyes developed even higher degrees of hyperopia.

Figure 5.

Spherical-equivalent, spectacle-plane refractive corrections plotted as a function of age for the treated (filled symbols) and fellow eyes (open symbols) of representative lens-reared controls (top row) and red-light-reared monkeys (bottom row). All of the animals were reared with +3.0 D lenses in front of their treated eyes and plano lenses in front of their fellow eyes. The thin gray lines in each plot represent data for the right eyes of the 39 normal control monkeys. In the top row the panels are arranged from left to right according to the maximum degree of hyperopic anisometropia; the monkeys represented in panels A and G exhibited the smallest and largest degrees of hyperopic anisometropia, respectively; the monkey represented in panel D exhibited the median degree of hyperopic anisometropia in the control group. In the bottom row, the panels are arranged from left to right in order of increasing hyperopia in the lens-reared eye.

The effects of the different rearing strategies on interocular differences in refractive error are summarized in Figure 6, which shows interocular differences in refractive error (treated eye – fellow eye) plotted as a function of age for individual NL (left column) and RL monkeys (right column). The diffuser- and lens-reared NL monkeys consistently developed significant interocular differences in refractive error that were outside the 95% confidence limits for NL control monkeys (see Table 1). In particular, at the end of the treatment period, 14 of 16 FD-NL monkeys and 16 of 18 −3D-NL monkeys had developed myopic anisometropias that were more than 2 SDs below the mean degree of anisometropia for NL control monkeys. All 9 +3D-NL monkeys developed compensating hyperopic anisometropias that were above the range of anisometropias observed in NL control monkeys. The +3D-RL monkeys also developed consistent hyperopic anisometropias (Fig 6F); however, the degree of anisometropia was significantly smaller than observed in the +3D-NL group (+1.16 ± 0.55 D vs +1.86 ± 0.54 D; T = 2.56; P = 0.01.) In contrast, the direction of anisometropia was not consistent in either the FD-RL (−0.46 ± 3.66 D) or the −3D -RL monkeys (−0.06 ± 0.95 D) and their average end-of-treatment anisometropias were less myopic than that exhibited by the FD-NL (−4.51 ± 3.99 D; T = 2.37; P = 0.04) and −3D-NL monkeys (−2.15 ± 1.04 D; T = 4.54, P = 0.001), but not different from that in NL control monkeys (−0.01 ± 0.28 D; P = 0.76 and 0.90).

Figure 6.

Interocular differences in refractive error (treated eye – fellow eye) plotted as a function of age for individual FD (A), FD-RL (B), −3D-NL (C), −3D-RL (D), +3D-NL (E) and +3D-RL monkeys (F). The large symbols on the right in each plot show the mean (±1 SD) anisometropia at the end of the lens-rearing period. The shaded area in each plot represents ±2 SDs of the mean anisometropia for the 39 normal control monkeys.

The consistency of the hyperopic shifts produced by the ambient red lighting is evident in Figure 7, which shows end-of-treatment refractive errors for the treated (filled symbols) and fellow eyes (open symbols) of all of the lens- and diffuser-reared NL monkeys and all the RL monkeys. For reference, the ametropias for both eyes of age-matched NL control monkeys are represented by open diamonds. To increase the visibility of the data, the results for the form-deprived monkeys are shown in a separate plot with a compressed ordinate scale. In comparison to the NL control monkeys, the median ametropias for both eyes of the monkeys in all RL treatment groups were significantly more hyperopic (see Table 1; P = 0.0001 to 0.02). The median ametropias for the fellow eyes of the lens- and diffuser-reared RL monkeys were also significantly more hyperopic than those of the fellow eyes in the lens- and diffuser-reared NL monkeys (P = 0.026 to 0.033). As we have observed previously in monocular lens-reared monkeys housed under white light, the fellow eyes of the −3D-RL (P = 0.05), +3D-RL (P = 0.10) and the FD-RL monkeys (P = 0.19) were marginally more hyperopic than the binocular RL controls, possibly reflecting differences in accommodative behavior associated with monocular fixation patterns (Smith III, Hung, Kee & Qiao, 2002). In addition, the median treated-eye refractive errors of the −3D-RL (P = 0.003) and the FD-RL monkeys (P = 0.001) were significantly more hyperopic than the treated eyes of their respective NL controls. There was a trend for the treated eyes of the +3D- RL monkeys (P = 0.10) to exhibit more hyperopic ametropias than the treated eyes of the +3D-NL controls.

Figure 7.

A. Ametropias obtained at ages corresponding to the end of the lens-rearing period for individual normal monkeys (open diamonds), red-light-reared controls (red diamonds), −3D-NL controls (black and white circles), −3D-RL animals (red and white circles), +3D-NL controls (black and white triangles) and +3D-RL monkeys (red and white triangles). B. End of treatment ametropias for normal controls (open diamonds), FD controls (black and white triangles), and FD-RL monkeys (red and white triangles) For the lens- and diffuser-reared monkeys, the treated and fellow eyes are represented by the filled an open symbols, respectively.

These relative hyperopic ametropias came about because the red ambient lighting altered the direction of refractive-error changes. In Figure 8, the relative changes in refractive error that took place during the treatment period are shown for the fellow or left eyes (filled symbols) of the NL lens- and diffuser-reared monkeys and all RL monkeys. For reference, the median changes in refractive error that were associated with emmetropization in NL control monkeys are represented by open squares in each plot. The longitudinal changes in refractive error that occurred in the fellow eyes of the lens- and diffuser-reared NL monkeys were not different from those observed in NL control monkeys (P = 0.12 to 0.74). Throughout the observation period the majority of the fellow eyes in the lens- and diffuser-reared NL monkeys exhibited slow relative reductions in hyperopia that overlapped the 25% to 75% confidence intervals for the NL control monkeys (shaded area). On the other hand, the longitudinal changes in refractive error for the fellow eyes of the RL monkeys were significantly different from NL control monkeys (P = 0.01 to 0.0002) and their respective lens- and diffuser-reared NL monkeys (P = 0.04 to 0.004). In particular, the RL monkeys typically exhibited increases in hyperopia, with only one animal in each of the +3D-RL and the FD-RL groups showing decreases in hyperopia that were comparable to the median for the NL control monkeys.

Figure 8.

Relative changes in spherical-equivalent refractive error plotted as a function of age for the control eyes of individual red-light monkeys (red symbols) and individual lens- and diffuser-reared controls (black symbols) The data were normalized to the refractive errors at the start of the treatment period (the first symbol in each function). The open squares represent the median changes in refractive error for age-matched normal control animals. The stippled area demarcates the 25% and 75% limits for the normal animals.

3.3. Effects of Ambient Red Lighting on Ocular Components

At the end of the treatment period there no significant interocular differences in corneal power, anterior chamber depth or lens thickness in any of the NL or RL groups (Table 1). In addition, there were no significant differences in the average corneal powers, anterior chamber depths, and lens thicknesses between NL control monkeys and those for any of the NL or RL treatment groups (one-way ANOVA, lens- or diffuser-reared eyes: F = 0.58 to 2.35 P = 0.06 to 0.43; fellow control eyes: F = 1.00 to 1.95; P = 0.11 to 0.41).

However, the refractive errors noted above were associated with alterations in vitreous chamber elongation. For example, as illustrated in Figure 9A, which shows the changes in vitreous chamber depth that took place during the treatment period, the monkeys that developed relative myopic anisometropias showed significantly larger increases in vitreous chamber depth in their treated eyes than in their fellow eyes (e.g., FD-NL, P < 0.001 and −3D-NL monkeys, P < 0.001). On the other hand, the monkeys that manifest relative hyperopic anisometropias exhibited significantly smaller increases in vitreous chamber depth in their treated versus their fellow eyes (e.g., +3D-NL monkeys, P < 0.001 and +3D-RL monkeys, P = 0.007). As a consequence, at the end of the treatment period, all subject groups with significant anisometropias had significant interocular differences in vitreous chamber depth (Table 1). The three RL-treatment groups that on average remained isometropic had similar increases in vitreous chamber depth in their two eyes (RL controls, P = 0.24; −3D-RL, P = 0.27; and FD-RL, P = 0.73).

Figure 9.

A. Changes in vitreous chamber depth over the course of the observation period for individual normal monkeys (open diamonds), red-light-reared controls (red diamonds), −3D-NL controls (black and white circles), −3D-RL animals (red and white circles), +3D-NL controls (black and white up triangles), +3D-RL monkeys (red and white up triangles), FD controls (black and white down triangles), and FD-RL monkeys (red and white down triangles) For the lens- and diffuser-reared monkeys, the treated and fellow eyes are represented by the filled and open symbols, respectively. B. End-of-treatment ametropias plotted as a function of the vitreous chamber depth / corneal radius ratio for individual red-light-reared animals. Only the right eye data are shown for the red-light controls. The open symbols represent the fellow eyes of the treated red-light monkeys.

The absolute hyperopic refractive errors observed in all RL-treatment groups at the end of the observation period were axial in nature, specifically due to reductions in vitreous chamber elongation. Figure 9B shows the association between the end-of-treatment ametropias and the ratio of vitreous chamber depth and corneal radius for all of the RL monkeys. There was a strong significant correlation between refractive error and the vitreous chamber/corneal radius ratio. This correlation is stronger than that between refractive error and absolute vitreous chamber depth. However, the vitreous chamber/corneal radius ratio can provide a truer measure of the significance of the vitreous chamber alterations. In particular, when there are no between group or between eye differences in corneal power, this ratio takes into account the effect of corneal power on absolute vitreous chamber depth and reduces the effects of normal between subject differences in corneal power on overall axial elongation.

It is well established that visual signals that alter the course of emmetropization in primates can also alter choroidal thickness, in particular imposed myopic defocus typically produces choroidal thickening (Hung, Wallman & Smith III, 2000, Troilo, Nickla & Wildsoet, 2000). As illustrated in Figure 10, which shows the average longitudinal changes in choroidal thickness for the 4 RL groups, the red light regimen produced qualitatively similar increases in choroidal thickness. The eyes of age-matched NL control monkeys (grey squares) exhibited modest increases in choroidal thickness over the course of the observations period (an average increase of 8.22 ± 10.31 µm). In comparison both eyes of the RL monkeys in every group exhibited marked increases in choroidal thickness that were sustained throughout the treatment period, reaching increases between 40 and 60 μm by the end of treatment. With the exception of the two- and four-week data obtained for the +3D-RL monkeys, the increases in choroidal thickness were well matched in the two eyes of a given animal.

For individual RL animals, the changes in refractive error were significantly correlated with the changes in choroidal thickness (Figure 10E; P < 0.002). However, unlike the changes in vitreous chamber described above, even the largest increases in choroidal thickness contributed only minimally to the absolute changes in refractive error. Calculations based on schematic eye models indicate that a 50 µm increase in choroidal thickness would produce about a 0.30 D hyperopic shift in refractive error at 150 days of age.

4.0. Discussion

The most consistent finding in this study was that rearing infant monkeys under narrow-band long-wavelength lighting resulted in progressive hyperopic ametropias. As a consequence, at the end of the rearing period, the fellow eyes for animals in all four RL treatment groups were significantly more hyperopic than age-matched normal monkeys reared under white fluorescent lighting. These hyperopic ametropias reflected absolute increases in hyperopia, not simply an interruption in the reduction in hyperopia that is normally associated with emmetropization. Similarly, the treated eyes of the animals in the diffuser-reared RL group and both lens-reared RL groups also developed significant degrees of hyperopia. In this respect, it is noteworthy that the long-wavelength ambient lighting not only prevented both FDM and defocus-induced myopia, but also augmented the hyperopia produced by imposed relative myopic defocus. All of the relative hyperopic ametropias produced by the red-light regimen were axial in nature and came about as a result of reductions in vitreous chamber elongation rate.

4.1. Between study comparisons of the effects of the spectral composition of ambient light on emmetropization.

There is growing evidence that manipulations of the spectral composition of ambient lighting can alter normal ocular growth and refractive development. Table 2 lists studies in which young chickens (Foulds et al., 2013, Rohrer et al., 1992, Seidemann & Schaeffel, 2002), guinea pigs (Jiang et al., 2014, Liu et al., 2011, Long et al., 2009, Wang, Zhaou, Lu & Chu, 2011), tree shrews (Gawne et al., 2017a, Gawne, Ward & Norton, 2017b) and monkeys (Liu. et al., 2014, Smith III et al., 2015) were reared with unrestricted vision under quasi-monochromatic lighting. In the majority of these studies, the animals were exposed to either relatively short- and/or relatively long-wavelength lighting. Although there were clear between-study differences in the intensity and spectral characteristics of the ambient lighting and the age of onset and duration of the rearing period, 11 of the 12 studies reported significant alterations in refractive development for at least one of their quasi-monochromatic lighting regimens. The Rohrer et al. (1992) study, which employed wavelengths from the extreme ends of the spectrum (UV <420 nm and deep red >650 nm) and low lighting levels, was the only study that failed to find any significant alterations in emmetropization.

Table 2.

Between study comparisons of the effects of the spectral composition of ambient lighting on emmetropization.

Reference	Species	Wavelength (nm)	Intensity Lux (human)@ (W/m2)	Duration	Result
Rohrer et al., 1992	Chicken	383 ± 12	0.002 lux (0.066 W/m2)	4 to 17 days	No effect
	Chicken	LW pass (50% at 665 nm)	1.52 lux (0.048 W/m2)	4 to 17 days	No effect
Seidemann & Schaeffel, 2002	Chicken	430 ± 7.5	5 lux	7 to 9 days	Hyperopic Shift
	Chicken	615 ± 7.5	5 lux	7 to 9 days	Myopic Shift
Foulds et al., 2013	Chicken	477 ± 13	33.4 cd/m2	1 to 15 or 43 days	Axial Hyperopia
	Chicken	641 ± 10	33.4 cd/m2	1 to 15 or 43 days	Axial Myopia
Liu et al., 2011	Guinea Pig	430 ± 13	3.6 – 5.9 lux (0.46 – 0.74 W/m2)	14 to 98 or 154 days	Axial Hyperopia
	Guinea Pig	530 ± 10	435 lux (0.75 W/m2)	14 to 98 or 154 days	Axial Myopia
Jiang et al., 2014	Guinea Pig	470 ± 5	50 lux	21 to 49 days	No effect
	Guinea Pig	600 ± 5	300 lux	21 to 49 days	Axial Myopia
Wang et al., 2011	Guinea Pig	480 ± 10	44 lux (0.46 W/m2)1	10 to 20 days	No effect
	Guinea Pig	530 ± 15	618 lux (1.05 W/m2)1	10 to 20 days	Axial myopia
Long et al., 2009	Guinea Pig	760 ± 7.5	150 lux	0 to 28 days	Axial Myopia
Gawne et al., 2017a	Tree Shrew	464 ± 10	600 lux	11 to 24 days#	Axial Myopia*
	Tree Shrew	628 ± 10	527 lux	11 to 24#	Axial Hyperopia
Gawne et al., 2017b	Tree Shrew	636 ± 10	527 lux	35 to 48 or 95# to 108 days	Axial Hyperopia
Liu et al., 2014	Monkey	455 ± 12.5	47 lux (1.4 W/m2)	50 to 407 days	No effect
	Monkey	610 ± 10	148 lux (0.43 W/m2)	50 to 407 days	Relative myopia
Smith et al., 2015	Monkey	LW pass (50% at 660 nm)	60 lux	25 to 146 days	Axial Hyperopia
Hung et al.	Monkey	630 ± 10	274 lux	24 to 151 days	Axial Hyperopia

^@Lux value calculated from the specified irradiance assuming all the power was at the peak wavelength

¹The irradiance units specified in the paper was W/cm²; we believe it should have been W/m²

^*more apparent when the blue lighted was flickered.

^#Days of visual experience

Because spectrally narrow-band lighting greatly reduces the potential chromatic cues associated with LCA, these results support the idea that the emmetropization process relies on chromatic signals to guide ocular growth (Rucker & Wallman, 2008, Rucker & Wallman, 2009). However, in this respect, it is very interesting that there were interspecies differences in the direction of the refractive changes produced by long-versus short-wavelength environments. In particular, in chickens and guinea pigs short-wavelength lighting consistently produced relative hyperopic shifts in refraction and long-wavelength lighting produced relative myopic shifts (see Table 2). This pattern of results was found over a wide range of lighting intensities. In some of these studies that employed short duration treatment periods (e.g., Seidemann & Schaeffel, 2002 and Wang et al., 2011), the magnitude of the changes in refractive error was comparable to the amount of LCA associated with the peak wavelengths of the ambient lighting. This quantitative agreement has been taken as evidence that under quasi-monochromatic lighting, the emmetropization process alters growth to maximize luminance contrast associated with the different focal points associated with LCA and that chromatic cues are not essential for normal emmetropization. However, with longer observation periods, the wavelength-dependent shifts in refractive error proved to be progressive in nature and continued to increase well beyond predictions based on LCA (Foulds et al., 2013, Liu et al., 2011).

In contrast to the pattern of results observed in chickens and guinea pigs, narrow-band lighting either failed to alter emmetropization in tree shrews and monkeys or induced refractive-error changes that were in the opposite direction. For example, rearing tree shews under short-wavelength lighting resulted in axial myopia; long-wavelength lighting consistently produced axial hyperopia (Gawne et al., 2017b). In monkeys, Liu et al. (2014) reported that short-wavelength lighting had no effect on refractive development, but that long-wavelength lighting produced relative myopia. However, there was substantial individual variability in refractive development in their long-wavelength group and there were no significant alterations in vitreous chamber depth. In our previous study, rearing infant monkeys with long-wavelength-pass filters consistently produced axial hyperopia that was qualitatively and quantitatively similar to that observed in our RL control animals (Smith III et al., 2015). In this study, the refractive changes observed over the course of the treatment period were more hyperopic than the median value for NL control monkeys in 25 of the 27 control eyes in RL-reared monkeys. The different refractive changes found in chickens and guinea pigs versus tree shrews and monkeys, particularly for long-wavelength lighting, does not appear to be related to the exact spectral composition or the intensity of the ambient lighting. The wavelengths and intensities in the chicken and guinea pig studies overlapped those employed in the studies involving tree shrews and monkeys (see Table 2). As noted in the Introduction, color vision in primates evolved differently from that in birds and many other mammals (Jacobs, 2009), which may have contributed to the observed inter-species differences. In this respect, like many mammals, tree shrews are dichromatic. However, tree shrews are considered to be closely related to primates (Luckett, 1980) and possibly the vision-dependent mechanisms controlling emmetropization in tree shrews evolved in a manner similar to that in primates.

Why did Liu et al. (2014) fail to observe hyperopic shifts in monkeys comparable to those observed in this study and in our earlier study involving long-wavelength ambient lighting? We had previously speculated that the differences between our results and those of Liu et al. (2014) might reflect absolute differences in lighting levels; Liu et al’s lighting levels were about 3–4 times higher than the lighting levels obtained through the long-wavelength-pass filters used in our earlier study (Smith III et al., 2015). In this respect, Rucker and Wallman (2008) have shown that low luminance levels reduce the ability of the emmetropization process to utilize luminance cues, thus increasing the likelihood that chromatic cues could influence ocular growth. However, in this study our lighting levels were about 1.8 times higher than those employed by Liu et al. (2014) and our long-wavelength lighting consistently produced hyperopia in infant monkeys. There were several other methodological factors that may have contributed to these between study differences. For example in the Liu et al. study, the peak emission of the long-wavelength lighting was slightly lower than that employed in our studies (e.g., 610 nm vs 630 nm), the monkeys were slightly older than ours at the onset of the treatment period (50 vs 25 days of age), and multiple monkeys were housed in individual cages that greatly restricted viewing distance (<1 m vs >4 M in our vivarium). This last difference may be significant because rearing adolescent monkeys (1 to 6 years of age) with restricted viewing distances (<50 cm) has been reported to promote the development of myopia. For example, for 1–2 year-old monkeys, Young found a median myopic shift of 2 D over a 12-month period (Young, 1961, Young, 1963)(for a review see (Smith III, 1998)).

4.2. Between study comparisons of the effects of the spectral composition of ambient light on compensation for imposed defocus.

The effects of narrow-band ambient lighting on compensation for imposed defocused also appears to vary greatly between species (see Table 3). Although more complete compensating ocular growth tends to occur in white light (Rucker & Osorio, 2008, Rucker & Wallman, 2008), young chickens exhibit axial refractive compensation for both myopic and hyperopic defocus when reared under narrow-band short -, middle-or long-wavelength ambient lighting (Rucker & Wallman, 2008, Schaeffel & Howland, 1991, Seidemann & Schaeffel, 2002, Wildsoet et al., 1993). Only chickens reared under ultraviolet lighting failed to exhibit lens compensation (Rohrer et al., 1992). The available results from tree shrews are qualitatively similar to those from chickens. Specifically, when reared under red ambient lighting, tree shrews develop myopic anisometropias in response to imposed monocular hyperopic defocus (Ward, Norton & Gawne, 2017). Both the treated and fellow eyes of these tree shrews exhibited relative hyperopic shifts and the compensating myopic anisometropias were larger than those observed in white-light controls. In contrast, rearing guinea pigs under short-wavelength lighting prevented compensation to imposed hyperopic defocus, while long-wavelength lighting suppressed compensating hyperopia in response to positive-lens-induced defocus. Instead refractive development in both the treated and fellow eyes of guinea pigs developed axial hyperopia in short-wavelength lighting and axial myopia in long-wavelength lighting (Jiang et al., 2014).

Table 3.

Between study comparisons of the effects of the spectral composition of ambient lighting on the phenomenon of lens compensation.

Reference	Species	Wavelength (nm)	Intensity Lux (human) (W/m2)@	Manipulation	Duration	Result
Schaeffel & Howland, 1991	Chicken	589	128 lux (0.25)	± 4 D lens	4 to 15 days	Axial compensation
Wildsoet et al., 1993	Chicken	550 or 589	33 lux (0.049) or 133 lux (0.26)	Recovery from FDM	14 to 42 days	Axial recovery
Rohrer et al., 1992	Chicken	383 ± 12	0.002 lux (0.066)	± 4 D lenses	4 to 17 days	no effect
	Chicken	LW pass (50% at 665 nm)	1.52 lux (0.048)	± 4 D lenses	4 to 17 days	Axial compensation
Rucker & Wallman, 2008	Chicken	460 ± 5	0.2 lux (0.0049)	+6 D or – 8 D lenses	3 days	Compensation due to AL
	Chicken	620 ± 5	0.47 lux (0.0018)	+6.0 D or – 8 D lenses	3 days	Compensation due to CT
Jiang et al., 2014	Guinea Pig	470 ± 5	50 lux	−4 D lenses	21 to 49 days	Axial Hyperopia; no lens effect
	Guinea Pig	600 ± 5	300 lux	+4 D lenses	21 to 49 days	Axial Myopia; no lens effect
Ward et al., 2017	Tree Shrew	628 ± 10	1000 lux	−5 D or FDM	11 to 13 or 23 days	Axial myopia in the treated eyes
Hung et al	Monkey	630 ± 10	274 lux	−3 D or FDM	24 to 151 days	No compensation; axial hyperopia
	Monkey	630 ± 10	274 lux	+3 D	24 to 151 days	Compensating axial hyperopia

^@Lux value calculated from the specified irradiance assuming all the power was at the peak wavelength

^#Days of visual experience

The pattern of results obtained in infant monkeys appears to be quite different from that observed in other commonly studied animals. In particular, infant monkeys reared with monocular negative lenses and exposed to long-wavelength lighting exhibited hyperopic shifts in both their treated and fellow eyes, but no anisometropic compensation. Monkeys reared with monocular positive lenses also show relative hyperopic shifts in both their treated and fellow eyes, but consistently developed compensating hyperopic anisometropias. Thus, in monkeys long-wavelength lighting effectively blocks defocus-induced myopia (and FDM), but not defocus-induced hyperopia. Based on the available data, there is no obvious explanation for the between study differences in the effects of quasi-monochromatic light on lens compensation other than potential idiosyncratic, inter-species differences.

4.3. Chromatic signals and emmetropization in primates.

The alterations in normal refractive development and in the responses to imposed defocus produced by narrow-band long-wavelength lighting indicate that chromatic signals play a major role in regulating emmetropization in monkeys. In some species, in the absence of chromatic signals, emmetropization for both hyperopic and myopic errors can be effectively regulated by maximizing luminance contrast signals (Schaeffel & Howland, 1991, Seidemann & Schaeffel, 2002, Wildsoet et al., 1993). However, at least with respect to reducing or eliminating hyperopia, that does not appear to be the case in monkeys. In particular the progressive hyperopic shifts observed in the RL-reared monkeys produced increasing levels of hyperopic defocus and concomitant reductions in luminance contrast. Similarly, the hyperopic shifts observed in eyes with imposed hyperopic defocus provide strong evidence that luminance contrast signals do not necessarily dominate emmetropization in monkeys. Interestingly, the compensating anisometropias observed in the +3D-RL monkeys indicate that luminance contrast signals are effective in guiding emmetropization for myopic errors in monkeys.

There is growing evidence that chromatic signals, like those associated with LCA, provide cues that decode the sign of optical defocus (Rucker, 2013, Rucker & Osorio, 2008, Rucker & Wallman, 2008). For example, comparisons of cone excitation levels or cone contrast signals between short- versus long-wavelength cone mechanisms can be used to determine if an eye is myopic or hyperopic (Rucker, 2013). Specifically, imposing or increasing myopic defocus would increase relative cone contrasts in long-versus short-wavelength cone mechanisms. As a consequence, rearing animals under quasi-monochromatic light could both eliminate useful cues and at the same time potentially generate anomalous directional cues for emmetropization. For instance, with our RL regime absolute excitation levels would be much higher in long- versus short-wavelength-cone mechanisms, a situation that would occur in a myopic eye in white light.

The results obtained from our lens-reared RL monkeys indicate that our RL regime masked luminance direction cues associated with hyperopic defocus and/or created erroneous direction cues normally associated with myopic defocus. Because our form deprivation regimen virtually eliminated luminance contrast signals associated with the sign of defocus in both NL and RL monkeys, the results obtained from our RL-FD monkeys suggest that our RL-rearing strategy produced erroneous direction cues. Specifically, the fact that our FD-RL monkeys developed hyperopic errors in both their treated and fellow eyes argues that our RL regime produced chromatic direction signals normally associated with myopic defocus. The progressive nature of the hyperopic shifts observed in untreated and fellow control eyes of RL-reared monkeys is in agreement with the idea that our RL regiment produced constant anomalous directional signals.

The ametropias in RL-reared monkeys appear to reflect the action of the vision- dependent mechanisms that normally regulate ocular growth. The constellation of ocular component changes associated with the hyperopic shifts in the untreated and fellow control eyes of RL-reared monkeys were identical to those associated with hyperopia produced by imposed myopic defocus in monkeys reared under broad-band white fluorescent lighting (Hung et al., 2018, Hung et al., 2000, Smith III & Hung, 1999). In particular, the observation that our RL regimen produced sustained increases in choroidal thickness suggests that the long-wavelength lighting produced a consistent signal to slow axial elongation within the vision-dependent cascade that normally regulates emmetropization.

4.4. Pupil mydriasis and melanopsin signaling under narrow-band long-wavelength lighting

All of our RL-reared monkeys developed pupil mydriasis several minutes after being placed in the RL environment. The pupil mydriasis was substantial, consistently outside the range of pupil sizes for animals housed under similar illuminances of white light, and was maintained throughout the treatment period. These pupil alterations were potentially the result of a decrease in melanopsin signaling to intrinsically photosensitive`retinal ganglion cells (ipRGCs).

Endogenous melanopsin-based phototransduction dominates the sustained light input to ipRGCs and plays a key role in maintaining pupil size during the day and at high lighting levels (Keenan, Rupp, Ross, Somasundaram, Hiriyanna, Wu, Badea, Robinson, Lowell & Hattar, 2016). As illustrated in Figure 1, there was little overlap of theabsorption spectrum for melanopsin and the emission spectra for the red LEDs. Thus, although the red light regimen would readily stimulate the middle- and long-wavelength-sensitive cones that also provide photoreceptive inputs to ipRGCs, the inputs derived from direct melanopsin photoreception would be expected to be very low.

The larger pupil sizes observed in the RL monkeys could potentially influence normal emmetropization and compensation for optically imposed defocus in several ways. The larger pupil sizes would be expected to decrease the depth of field and increase the size of retinal blur circles produced by an out-of-focus image. Both of these changes would potentially increase the strength of control signals associated with defocus, presumably increasing the effectiveness of a given dioptric error in altering ocular growth. In addition, the increase in pupil size would potentially alter the profile of monochromatic high-order aberrations. In particular, for a 5-mm pupil, 3-week-old monkeys exhibit higher amounts of positive spherical aberration (~ +0.20 microns) that decrease to normal adult levels (~ −0.01 micron) by about 100 days of age (Ramamirtham, Kee, Hung, Qiao-Grider, Roorda & Smith III, 2006). Thus, early in the rearing period, positive spherical aberration could potentially reduce the strength of visual signals produced by hyperopic defocus and increase the strength of signals associated with myopic defocus (Applegate, Marsack, Ramos & Sarver, 2003). However, variations in pupil size would have little effect on the spatial characteristics of the retinal image in form-deprived eyes. The fact that the treated eyes of the FD-RL monkeys also exhibited increases in choroidal thickness and hyperopic shifts that were comparable to those observed in the fellow control eyes of other RL monkeys and the treated eyes of −3D-RL monkeys suggests that the observed increases in pupil diameter did not contribute in a significant manner to the obvious alterations in ocular growth and refractive development produced by red light.

Based on their melanopsin-mediated sensitivity to short-wavelengths of light that are prevalent in outdoor scenes and their synaptic interactions with retinal neurons known to be involved in some aspects of vision-dependent ocular growth, it has been suggested that ipRGCs may play a role in the protective effect of time outdoors on the onset myopia in children (Beckett and Mutti, 2017; Optometry and Vision Science, program #175384). If melanopsin signaling in ipRGCs plays a key role in preventing myopia, it would be expected that our RL-regimen, which appears to have greatly reduced the sustained light signals to ipRGC, should have resulted in relative myopic refractive changes. However, we observed the opposite response in our RL controls. In addition, the protective effects of red light on lens-induced myopia and FDM in monkeys demonstrate that melanopsin signaling to ipRGCs is not necessary to prevent vision-dependent myopia, nor does melanopsin signaling appear to be essential to produce hyperopic shifts in refractive error.

Melanopsin signaling via the ipRGCs plays a role in maintaining ocular diurnal rhythms, some of which are associated with the visual regulation of ocular growth (Nickla, Wildsoet & Wallman, 1998). In this respect, it is likely that our RL regimen altered, at least initially, some of these diurnal rhythms. Measures of physical activity levels over the 24-hour daily light-dark cycle showed that, over the first 40 days of treatment, our RL monkeys were more active at night than NL controls (Arumugam, Hung, Ostrin, She, and Smith III, 2018; ARVO abstract #5043). Although sleep patterns returned to normal before the end of the treatment period, it is possible that some ocular rhythms were altered throughout the observation period. Wang et al. (Wang et al., 2011) have shown in guinea pigs that, in comparison to narrow-band blue light (480 nm), green light (530 nm) suppresses circulating melatonin levels and the expression of retinal melanopsin mRNA and protein. Our RL regimen very likely produced a similar pattern of changes in our monkeys. Nevertheless, the RL regimen resulted in hyperopic shifts rather than myopic shifts. Thus, it seems unlikely that any alterations in diurnal rhythms produced by the RL regimen and the accompanying alterations in melanopsin signaling contributed directly to the observed hyperopic changes.

4.5. Implications for treatment strategies for myopia.

Given the extreme nature of the lighting manipulations that are typically employed in animal studies, one has to be very cautious extrapolating the results from these studies to humans. Although it may be feasible to limit human exposure to certain wavelengths with broad-band filters, electronic displays or lighting systems, it is not reasonable or feasible to impose narrow-band ambient lighting on humans comparable to those used in animal research, especially for the entire daily lighting cycle. Nonetheless, our results from monkeys have some implications for strategies that have been proposed as potential treatment options to reduce the progression of myopia in humans.

Several observations support the hypothesis that reducing the amount of long-wavelength light and/or increasing the amount of short-wavelength light in the environment would reduce the risk for myopia progression in children. For example, it has been argued that increasing the amount of short-wavelength light would reduce accommodative efforts for near targets, potentially reducing accommodative lag and the resulting hyperopic defocus, i.e., reducing a strong stimulus for myopic growth (Seidemann & Schaeffel, 2002). It has been proposed that having children read from paper that preferentially absorbs longer wavelengths or by viewing through blue filters could have therapeutic benefit against myopia (Kroger & Binder, 2000). It has also been speculated that the predominance of short-wavelength light in outdoor environments contributes to the protective effects of time outdoors against myopia in children, possibly via the forward movement of the eye’s optimal focal plane as a consequence of LCA, by influencing the relative photon catch by long- versus short-wavelength sensitive cones, or by altering the location of excitation produced by light absorption along the photoreceptor outer segment (Foulds et al., 2013). In this respect, Torii et al (2017) reported that exposure to violet light (360 to 400 nm) suppressed myopia progression (Torii, Kurihara, Seko, Negishi, Ohnuma, Inaba, Kawashima, Jiang, Kondo, Miyauschi, Miwa, Katada, Mori, Kato, Tsubota, Goto, Oda, Hatori & Tsubota, 2017). Although Torii et al.’s results have been questioned (Schaeffel & Smith III, 2017), it has been suggested that efforts should be made to increase the amount of short-wavelengths in artificial indoor lighting.

Although these concepts are supported in some sense by the results from chickens and guinea pigs (see Table 2), our observations in monkeys and those in tree shrews (Gawne et al., 2017a, Gawne et al., 2017b) would suggest the exact opposite; to reduce the risk for myopia or myopia progression, efforts should be made to increase the prevalence of long-wavelengths in the environment. However, the diversity of the results summarized in Tables 2 and 3 emphasizes that our understanding of the effects of the spectral composition of ambient lighting on ocular growth and refractive development is still rudimentary, especially in primates. At the present time, it does not seem reasonable to propose treatment strategies for myopia that are based on manipulations of the spectral composition of ambient lighting. Nonetheless, the fact that manipulating the wavelength of ambient lighting can clearly alter normal refractive development in monkeys and prevent myopia in response to form deprivation and imposed hyperopic defocus warrants the systematic investigation of the potential role of the spectral composition of ambient lighting in the genesis and treatment of common refractive errors.

Highlights.

In infant monkeys, narrow-band, long-wavelength lighting:

• produces axial hyperopia

• prevents form-deprivation myopia

• retards myopic compensation to imposed hyperopic defocus

• augments hyperopia in response to imposed myopic defocus

Chromatic cues have powerful effects on ocular growth in primates