Myopia and the optics of the eye

The optics of the human eye is inferior to most commercial lenses which are made of glass and plastics. However, given the soft and rubbery material that it is made from, the optics is surprisingly good. In the fovea, the spatial resolution of the projected image has enough details to stimulate two adjacent photoreceptors differently - despite that they are only about two micrometers apart. In young people (who have better optics than older people), visual acuity correlates only weakly with the quality of their optics - indicating that optics itself is not limiting (Villegas et al 2008). The prize to be paid for good optics is a high geometrical precision. In emmetropia (normal-sightedness, sharp vision at distant objects with relaxed accommodation), the axial length of the eye is tuned to the focal length of cornea and lens with a precision of about a tenth of a millimetre. This is surprising, given that cornea, lens, and eye length all follow their own growth programs, so that the adult eye is by no means a scaled version of the juvenile eye. It becomes clear that refractive errors cannot be random deviations from emmetropia. Rather, a closed feedback loop must control axial eye growth to fine-tune axial length to the focal length. In the case of myopia, the error signal must deviate to provide the wrong information. But what do the signals look like and how are they generated? If this were known, the development of myopia could be stopped.

In humans, it is difficult to achieve any progress identifying possible error signals in retrospective studies because myopia is determined by a mixed genetic and environmental input. Typically, multifactorial regression analysis is performed to extract the influence of possible individual factors, but correlations do not tell us anything about mechanisms and it remains open whether or not any links are causal.

Genetic and environmental factors in myopia

The influence of genes is clearly indicated by the observation that myopia in children is partially predicted by the myopia of the parents. For instance, in the Orinda study (Mutti et al 2002), only 6% of the 14-year old children were myopic if neither of the parents were myopic, 18% if one parent was myopic and 33% if both were myopic. Also studies in twins demonstrate a high correlation in refractions (about 90% in monozygotic twins, and 60% in heterozygotic twins; i.e. Lopez et al 2009). Even intelligence (i.e. Saw et al 2004) or body height or weight (Dirani et al 2008) have been significantly correlated with myopia when large populations were analyzed. Another possible factor is the initial shape of the eye: more relative hyperopia in the periphery was related to more myopia development, as first described by Hoogerheide et al (1971). More recently, the role of peripheral refraction in myopia development has been studied by many laboratories (see below).

On the other hand, genetics cannot explain why the incidence of myopia doubled in eight year-old children in Taiwan from 1995 to 2005 (from about 25 to 50%, recently levelling off), or why myopia tripled in big cities in China in the past 30 years (Morgan and Rose 2005). Up to 90% of students are wearing contact lenses or spectacles. Also in the United States, myopia rose from about 25% in 1971-72 to about 42% in 1999-2004 (Vitale et al 2009). Thousands of years would be necessary to select for myopia-inducing genes that could explain such a change. Therefore, environmental factors, related to industrialization and urbanisation must carry the risk for myopia development.

Environmental factors that have been found in epidemiological studies

The list of possible environmental risk factors, determined from studies in children, is long: extensive near work and short reading distances (which may include extensive computer work; review Morgan & Rose 2005), undercorrection (i.e. Adler & Millodot 2006), soft contact lenses (i.e. Blacker et al 2009), night light (Quinn et al 1999), no smoking (i.e. Stone et al 2006), no breast feeding (Chong et al 2005), no sports (i.e. Jones et al 2007), no “outdoor activity” (i.e. Jones et al 2007 - Fig. 1; Rose et al 2008), high sugar or starch intake (Cordain et al 2002), stress and intense education (review in Morgan and Rose 2005), and many others. All of these factors showed significant associations with myopia in the studied populations. But which factors are the most important ones and worth better control in the future? Don Mutti (2008) had great doubt that it can be put as simply as “hanging around at the pool and smoking some cigarettes”. At the 2010 ARVO meeting (abstract #2968), it was claimed that outdoor activity only reduces the risk of myopia onset but has no major effect on its progression - which is not expected from experimental results in chickens (see below).

Fig. 1: The time spent outdoors correlates negatively with the incidence of myopia in children. However, the number of myopic parents is equally important. The problem of finding out which factors are most important in myopia development is that multiple factors are always involved, but also the long time delays until an effect can be seen (replotted after Jones et al 2007).

Lessons on myopia from animal models

Animal models of myopia offer the advantage that the genetic and environmental factors can be separated. Environmental factors can be intentionally varied while the genetic background is kept similar, or animals can be bred to select for higher susceptibility for myopia development to identify underlying genes. Both experiments have been undertaken. A few major results are described below, since some of them may shed light on myopia development in humans.

1- Coordinated eye growth requires that the image projected on the retina contains fine details and has good contrast. If the retina “sees” a degraded image, it typically triggers enhanced axial eye growth which leads to high amounts of myopia. This has first been shown in monkeys (Raviola and Wiesel 1977), then in chickens (Wallman et al 1987) and later in many other animal models (review: Wallman and Winawer 2004). Since this type of myopia develops in response to reduced vision, it is called “deprivation myopia”. It has been found in humans as well, in the case of infantile cataract, keratitis, or ptosis. Deprivation myopia develops only in the deprived eye, and even only in the retinal area that is deprived (first shown by Wallman et al 1987) - a strong indicator that the retina controls the growth of the underlying sclera directly. This finding was unexpected and challenges an older assumption that axial myopia is triggered by extensive accommodation and convergence, and perhaps high intraocular pressure.

2- If eye growth were controlled only by a (retinal) mechanism that can trigger deprivation myopia, it is unclear how the best match of focal length and eye length can ever be achieved during development. With a poor image on the retina, the eye should continue to grow in a positive feedforward manner until it becomes extremely myopic. That this is not observed suggests the action of a second, inhibitory mechanism. In 1987, it was found that defocus imposed by positive lenses inhibits axial eye growth in chickens (Schaeffel et al 1988) while negative lenses accelerate axial eye growth, similar to deprivation of sharp vision. However, a large difference exists since defocus can trigger the visual feedback loops for the control of axial eye growth in a “close-loop” fashion: the errors signal decline until the system can finally find the most appropriate refraction - typically emmetropia or slight hyperopia. Since neither accommodation nor an intact optic nerve are required for this bi-directional growth response (i.e. Wildsoet and Wallman 1995), it is clear that the mechanism must reside within the eye - starting in the retina and including RPE, choroid and sclera (Fig. 2).

Fig. 2: The retina in the back of the eye can detect the position of the plane of focus and inhibit or accelerate axial eye growth to achieve the best match between photoreceptor plane and image plane over time. The communication between retina and sclera is surprisingly direct: growth signals are released most likely from amacrine cells (A) diffuse or are transported through the retinal pigment epithelium and choroid (4) to finally reach the sclera (leftmost layer).

Experiments with spectacle lenses have been repeated in a number of animal models with similar results. Recently it was found that, even in monkeys, changes in eye growth can be induced locally by degrading the image quality with a hemifield diffuser in front of one eye (Smith et al 2009). The initial finding of Hoogerheide et al (1971) also triggered experiments in monkeys in which the periphery of the visual field was degraded or defocused. It was found that the foveal refraction was shifted despite that the central 37 deg of the visual field were unobstructed - indicating that emmetropization is not dependent on a foveal input.

Even though it is now clear that the retina controls axial eye growth in both directions - stimulation if the image focus is behind the retina (during negative lens wear) and inhibition if the plane of focus is in front of the retina (during positive lens wear) - it remains a major mystery how the sign of defocus is determined by the retina within a few minutes (Zhu et al 2005), also in the absence of accommodation and with only one viewing distance available (review: Wallman and Winawer 2004).

Recent experiments have shown that the “speed” of lens compen-sation is altered if chickens are exposed to bright light (15000 lux) for a few hours a day, in addition to regular laboratory illumination (500 lux, Ashby and Schaeffel 2010; Fig.3).

Fig. 3: Chicks are raised with 5 hours of exposure to bright light (15000 lux) per day. The spectrum of the movie lamps is shown in the bottom right, compared to sun light. Under this illumination, the chicks (here in black) develop only half the amount of myopia that they develop under normal laboratory illumination (500 lux).

Interestingly, while it takes the eye longer to compensate negative lenses, the compensation of positive lenses is accelerated. Obviously, the gains of the two antagonistic mechanisms are inversely affected by light - which could perhaps explain the beneficial effect of “outdoor activity” on myopia incidence in children.

3- Both deprivation myopia and negative lens-induced myopia can be suppressed by a number of drugs, dopamine agonists, neurotoxins against catecholamines, cholinergic antagonists, glucagon (in chickens) and others. In chickens, these drugs are typically applied by intravitreal injection - not a very attractive approach if one ever considers application in children. However, cholinergic antagonists, like atropine, are also effective if given as eye drops, in mice (Barathi et al 2009), guinea pigs and humans. On the other hand, some neuromodulators can also enhance myopia: as an example, insulin renders the retina insensitive to the sign of imposed defocus, and triggers the development of myopia with both positive and negative lenses (Feldkaemper et al 2009; Zhu and Wallman 2009).

What can be deduced for the development of myopia in children?

1- Obviously, children become myopic without any lenses being placed in front of their eyes - so what moves the focal plane behind the retina to stimulate eye growth? One hypothesis was raised by Gwiadza et al in 1993: since many subjects accommodate slightly too little during reading (the “lag of accommodation”), reading could shift the focal plane behind the retina, causing a similar condition as when weak negative lenses are worn. However, after many studies since 1993, it seems that there is only a weak link between both factors. The lag of accommodation is often not correlated to myopia progression (i.e. Mutti et al 2006; Weizhong et al 2008). On the other hand, in a large study in which children were treated with progressive addition lenses (PAL), rather than with single vision lenses, the progression of myopia was reduced with PAL, and the largest effect was seen in those children who had a pronounced lag of accommodation (Gwiazda et al 2004). A role of oculomotor functions in refractive development is therefore obvious - but it is not the lag alone.

2- The finding that bright light slows the development of myopia that is induced by deprivation of sharp vision, or by negative lenses in chickens, provides an explanation for the inverse correlation between outdoor activity and myopia in children. Perhaps exposure to bright light for some time each day is sufficient to retard the development of myopia. Experiments have now to be repeated in monkeys (or this has already been done) - and they will tell about the importance of the light factor in human myopia.

3- The observation that the peripheral refraction has a prominent effect on the refractive development in the fovea has important implications for the design of spectacle lenses. A recent study (Tabernero et al 2009; Fig.4) has shown that conventional spectacles to correct myopia induce variable amounts of hyperopia in the periphery of the visual field. Given that the retina controls growth in the back of the eye at each location, this condition should stimulate compensatory eye growth and drive myopia development. New spectacle designs are currently being tested which impose myopic, rather than hyperopic, defocus in the periphery (radial refractive gradient lenses, RRG lenses).

Fig. 4: Measurement of the peripheral refractive error in human subjects, using a newly developed scanning infrared photorefractor (Tabernero et al 2009). Using this device, it was found that conventional spectacle lenses induce various amount of peripheral hyperopia (green lines, bottom) but that a newly developed RRG lens induces some peripheral myopia (as intended).

4- The most potent known drug against myopia is atropine since it can stop eye growth in children completely during the first months of (daily) application. The well known side effects (pupil dilatation, paralysis of accommodation) require that reading glasses and perhaps also sun glasses are used. A larger disadvantage is however that the beneficial effects on myopia progression wear off over time. Atropine largely loses most of its effect in the third year of treatment and recent data suggest that eye growth is then even faster than in vehicle-treated control eye (Tong et al 2009). Still, it is claimed that end-point myopia is still less than in untreated control groups. Because of less severe effects on pupil size and accommodation, another muscarinic antagonist was introduced against myopia in 1991 (Stone et al 1991): pirenzepine. It is about half as potent as atropine, but since it had already been on the market to treat gastritis, and had gone through tolerability studies, it made it through phase 1 and phase 2 clinical studies with acceptable success (max. 50% inhibition of myopia). The problem is currently that the phase 3 study required by the FDA involves a 5 year trial with costs that seem currently out of range.

A few remarks on high myopia

In this summary features of the feedback loop that control axial eye growth based on visual input were described. Any feedback loop has a linear range of operation, in which the response is appropriate to minimize the error signal. This has been found also here: chicken eyes can compensate lens powers from perhaps -15 to +15 D, monkeys can do perhaps a fourth of this. And in humans, the possible range of compensation should be even less. Outside the regulated range, other (non-visual) factors may determine myopia development. A striking observation was that recovery from myopia (requiring “shrinkage of the eye”) is very rare - it has been observed only in two cases - in children treated with atropine every day for only the first months of treatment, and in chickens treated with positive lenses. It is not clear at which point the development of high myopia can still be influenced by visual factors. Typically, it starts early and involves a mechanically weakened scleral tissue which may not be able to resist against introcular pressure - but this is not well known. Myopia seems always to interfere with scleral tissue integrity since the eye balls even of low myopes are more irregular than in emmetropes (Tabernero and Schaeffel 2009). Several genetic loci for high myopia have been discovered (Hornbeak and Young 2009) but the genes themselves and the biochemical mechanisms in the sclera have not been fully worked out.