Ciliary Muscle, Eye Shape, and Accommodation in Adults with Anisometropia

Ciliary muscle, eye shape, and accommodation in adults with anisometropia

THESIS

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University

Mallory Kuhlmann Kuchem

Graduate Program in Vision Science

The Ohio State University

2012

Master's Examination Committee:

Melissa Bailey, OD, PhD, Advisor

Nick Fogt, OD, PhD

Donald Mutti, OD, PhD

Mallory Kuhlmann Kuchem

2012

Abstract

Purpose: The purpose of this study was to investigate the relationships between ciliary muscle thickness, refractive error, axial length, and accommodative lag in a sample of anisometropic adults.

Methods: The right and left eyes of 30 adult subjects were measured. All subjects possessed a difference of at least one diopter of refractive error between the two eyes.

Accommodative lag to a 4.00 D target was measured. Then, cycloplegic measurements of the nasal ciliary muscle thickness (CMT), spherical equivalent refractive error, and axial length were made. Ciliary muscle thickness measurements were made at the maximum ciliary muscle thickness (CMTMAX) and at 1.0 mm (CMT1), 2.0 mm (CMT2), and 3.0 mm (CMT3) posterior to the scleral spur using the Zeiss Visante™ Anterior Segment

OCT. Simple linear regression and multilevel regression models were used to compare ciliary muscle thickness, refractive error, axial length, and accommodative lag both across and within subjects.

Results: Across subjects, CMT was significantly negatively associated with mean refractive error and significantly positively associated with mean axial length when refractive error and axial length were first averaged between the two eyes of each subject.

This relationship was most pronounced for CMT2 (p < 0.0001). When CMT2 was controlled for, CMTMAX and CMT1, which represent the presumed radial/circular fibers ii of the ciliary muscle, are significantly positively associated with mean refractive error (p

< 0.0001) and significantly negatively associated with mean axial length (p = 0.0002) across subjects. Within subjects, there was no significant difference in ciliary muscle thickness. Accommodative lag was significantly negatively associated with the thickness of the ciliary muscle, but only at CMT2 (p = 0.02) and CMT3 (p = 0.01). Within subjects, no difference in accommodative lag between the two eyes was observed.

Conclusions: Across subjects, this sample of anisometropic subjects behaves in accordance with the literature: thicker ciliary muscles are associated with increased myopic refractive error and axial length. Within subjects, however, this relationship ceases to exist, indicating that an eye can grow longer and more myopic than its fellow eye without resulting in an increase in ciliary muscle thickness. Across subjects, the longer, more myopic eyes tended to have thicker posterior portions of their ciliary muscle

(posterior to CMT2) and lesser amounts of accommodative lag, while the shorter, more hyperopic eyes tended to have thicker anterior portions of their ciliary muscle and larger amounts of accommodative lag. This suggests a division in both structure and function at the position of CMT2. Specifically, the portion of the ciliary muscle posterior to CMT2 may be more implicated in sustaining accommodation and reducing lag, while the portion of the ciliary muscle anterior to CMT2 may play a greater role in accommodative workload.

iii

Dedication

This document is dedicated to my loving husband, Chad.

Acknowledgments

I would like to acknowledge the extensive knowledge and guidance provided by my advisor, Melissa Bailey, OD, PhD. This work would also not have been possible without the image analyses performed by Chiu-Yen Kao, PhD, and the statistical analyses performed by Loraine Sinnott, PhD. I would also like to acknowledge the frequent intellectual and technical advice provided by Don Mutti, OD, PhD throughout this endeavor. This project was supported in part by NIH grants T35 EY-007151, R24-

EY014792, and KL2 RR025754 from the National Center for Research Resources.

Vita

2004...... Olentangy High School

2008...... B.S. Biology, University of Virginia

Fields of Study

Major Field: Vision Science

Table of Contents

Abstract ...... ii

Dedication ...... iv

Acknowledgments...... v

Vita ...... vi

Fields of Study ...... vi

Table of Contents ...... vii

List of Tables ...... viii

List of Figures ...... ix

Introduction ...... 1

Methods...... 16

Results ...... 23

Discussion ...... 38

References ...... 48

vii

List of Tables

Table 1. Age, Spherical Equivalent Refractive Error, and Ciliary Muscle Thickness ..... 24

Table 2. The relationship between refractive error and the presumed circular/radial fibers at CMTMAX and CMT1 ...... 29

Table 3. The relationship between axial length and the presumed circular/radial fibers at

CMTMAX and CMT1 ...... 29

Table 4. Multilevel linear regression model for CMT and the more myopic eye within a person ...... 32

Table 5. Multilevel linear regression model for CMT and the longer eye within a person

...... 33

Table 6. Multilevel linear regression model comparing the presumed radial/circular fibers at CMTMAX and CMT1 between the two eyes of anisometropic subjects ...... 34

Table 7. Multilevel linear regression model comparing the presumed radial/circular fibers at CMTMAX and CMT1 between the two eyes of anisometropic subjects ...... 34

Table 8. The relationship between accommodative lag for a 4.0-D stimulus and Ciliary

Muscle Thickness...... 35

Table 9. T-tests of the hypothesis that the mean difference between the two anisometropic eyes of each subject was zero for multiple parameters ...... 37

viii

List of Figures

Figure 1. Accommodative stimulus-response function...... 12

Figure 2. Representative Visante™ image of the nasal ciliary body while the subject views an external fixation target...... 22

Figure 3. Mean ciliary muscle thickness plotted against mean spherical equivalent refractive error for each subject...... 26

Figure 4. Mean ciliary muscle thickness plotted against mean axial length for each subject...... 27

Figure 5. Presumed circular fibers at CMT1 plotted against the mean spherical equivalent refractive error for each subject...... 30

Figure 6. Presumed circular fibers at CMTMAX plotted against the mean spherical equivalent refractive error for each subject...... 30

Figure 7. Presumed circular fibers at CMT1 plotted against the mean axial length for each subject...... 31

Figure 8. Presumed circular fibers at CMTMAX plotted against the mean axial length for each subject...... 31

Introduction

The aim of this study was to investigate the relationships between ciliary muscle thickness, refractive error, axial length, and accommodative lag in a sample of adults with at least one diopter of anisometropia between the two eyes. A large body of research exists on the subject of myopic eye growth, but very few investigators have addressed the role of the ciliary muscle in myopia. Similarly, there is a widespread interest in the role of accommodation and accommodative lag in the field of myopia research, but the link between the structure of the ciliary muscle and its function in accommodation has not been adequately investigated. This study purposely recruited anisometropic subjects to examine differences in ciliary muscle thickness, refractive error, axial length, and accommodative lag that occur within the same person. Thus, we compared two eyes with different refractive error and axial length that were identical in genetic and environmental make up.

Myopia is a relevant research topic based on its rampant prevalence in every population and the significant public health burden it places on society. Myopia is a refractive condition in which the axial length of an eye exceeds the focal point formed by the refractive components of the eye, namely the cornea and crystalline lens. It is generally treated with either spectacle or contact lens correction, which costs billions of dollars every year in the United States alone.1 The prevalence of myopia in the United

States has increased from a reported 25% of people aged 12-54 years in 1971-19722, to almost 42% in 1999-2004.3-4 The cause of this statistically significant increase in myopia prevalence is unknown, as is the cause of myopia development itself.

Myopia can be thought of as a mismatch between the power of the refractive components and the length of the eye; the eye is either “too strong” in the case of refractive myopia or “too long” in the case of axial myopia. Studies have shown, however, that most often axial length is the culprit, not a refractive system that is too strong.5 In contrast, hyperopia is a mismatch in the opposite direction; the refractive components of the eye focus the light behind the retina, often because the axial length of the eye is “too short”.

In general, refractive error varies little between the two eyes of a person. When the refractive error of one eye differs from that of the other, the person is termed to have anisometropia. Estimates for the prevalence of anisometropia with ≥ 1.00 D difference between eyes range from just under 20% in a German/Austrian population of refractive surgery candidates6 to just under 7% in a population of Iranian nationals7. In the

German/Austrian sample, the majority of anisometropic subjects had only a mild amount of anisometropia (1.00 to 2.00 D) and the prevalence of anisometropia ≥ 3.00 D in the population was only 1.4%.6 The literature generally agrees that a difference of at least

1.00 D between the eyes is required to be classified as anisometropia, which guided our recruitment criteria for this study.

Although the state of myopia once it has developed is well understood, there is still much research and debate about the underlying mechanisms of myopic eye growth.

What factors contribute to an eye becoming myopic, and what specific changes in the physical ocular structure result in the elongated myopic eye? Ultimately, if the causative stimuli and the ocular mechanics of myopic eye growth can be determined, perhaps those mechanisms can be exploited in various treatment strategies to prevent or slow myopic eye growth.

Multiple studies have shown that the myopic globe is relatively more prolate than the emmetropic globe.8-10 This relative axial elongation accounts for the more myopic refractive error in the more prolate eyes. Different investigators have proposed differing theories as to how this prolate elongation occurs. Investigators in the Collaborative

Longitudinal Evaluation of Ethnicity and Refractive Error (CLEERE) Study proposed that the prolate nature of myopic eye growth may be a result of an internal source of growth restriction in the equatorial region of the globe.8 The crystalline lens as the source of this restriction was ruled out when the lens was shown to exhibit an extensive capacity to stretch during in vitro experiments.11 The ciliary muscle remains a potential source of this internal restriction.12 Atchison et al.9, however, suggested that it may be external constraints of the orbital wall that govern the prolate nature of myopic eye growth. This theory was rejected by the CLEERE Study Group, however, when they found that the onset of myopia is preceded by an acceleration of myopia progression, axial elongation, and relative peripheral hyperopia. The CLEERE investigators reasoned that this acceleration in rate would likely not occur if a steady external source such as the orbital walls were restricting equatorial growth.13 Logan et al.10 proposed that the axial and transverse dimensions of myopic eye growth may be regulated independently from one

3 another. The present study will further examine the role of the ciliary muscle as a possible physical source of internal growth restriction, as proposed by the CLEERE study group.8

The actions of the ciliary muscle are responsible for the changes in crystalline lens shape and position that govern the eye’s ability to change focus from distance objects to near objects. This change in focus is referred to as accommodation, and its mechanical forces are driven by the ciliary muscle. When the ciliary muscle contracts, the anterior portion thickens while the posterior portion thins.14 This causes the crystalline lens to thicken and move forward slightly, thus increasing the plus power of the lens and moving the position of the image forward toward the retina to bring closer objects into focus.15-16 The ciliary muscle is a worthy of study in the context of myopic eye growth because it has the potential for both physical and functional contribution to eye growth.

Physically, the ciliary muscle could potentially act as an internal restriction to globe growth, forcing any future eye growth to occur in the axial direction and thus causing the eye to become more prolate in shape and more myopic in refractive error. Functionally, the ciliary muscle could be implicated in accommodative effects on eye growth, as has been suggested17-19 and which will be discussed in depth later.

Three studies in the literature have shown a consistent relationship between ciliary muscle thickness and refractive error. Van-Alphen (1961) used the results of globe expansion experiments to predict that the ciliary muscle would thin with an increase in eye growth.5 Oliveira and co-workers (2005), however, used ultrasound biomicroscopy to show that ciliary muscle thickness in a sample of adult subjects increased with increasing

4 myopia and increasing axial length. 20 Bailey and colleagues (2008) used the Visante anterior segment OCT to show that myopic children had thicker ciliary muscles than non- myopic children.21 Muftuoglu and co-workers (2009) used ultrasound biomicroscopy to measure ciliary muscle thickness in adult subjects with unilateral high axial myopia and found that for most subjects, the ciliary body thickness in the more myopic eye was greater than the ciliary body thickness in the less myopic eye.22 Only the latter paper used a within-subject comparison to evaluate the relationship, yet the subjects had degrees of anisometropia not normally encountered in the general population.6

In the present study we will test for differences in ciliary muscle thickness and refractive error or axial length between eyes within the same subject possessing a moderate amount of axial anisometropia. Genetics has long been implicated as the predominant risk factor for myopic development23, so comparing eyes of subjects with anisometropia will allow us to control for differences across subjects, such as height, gender, and genetic makeup, for which we cannot account in a cross-sectional study. It is possible, for instance, that longer eyes simply have larger ciliary muscles, much in the same way that a tall man has much larger biceps than a toddler. Oliveira et al. alluded to this when he found that the major association was between ciliary muscle thickness and axial length, not refractive error.20 Utilizing an anisometropic study population allows us to filter out the possibility that the overall size of a person could influence her axial length and thus her ciliary muscle thickness. Any spurious correlations due to height, gender, and genetics will be avoided. The results of the current study of anisometropic eyes will assist us in determining if ciliary muscle thickness is directly correlated with

5 refractive error and axial length in individual eyes, and if it is possible for an eye to grow longer with axial myopia and have no change in ciliary muscle thickness.

Ciliary muscle thickness can be thought of as a measurement of the abundance of muscle fibers at the particular measurement location. The composition of the ciliary muscle, however, is not homogenous. The smooth muscle bundles throughout the ciliary muscle run in three different orientations; an outer longitudinal portion, an intermediate radial portion, and an inner circular portion. These portions are not distinct, but rather, form a functional syncytium that rearranges to effect the changes in ciliary muscle shape and position seen during accommodation. The longitudinal fibers originate in the epichoroid and insert anteriorly at the scleral spur. The radial fibers fan from the anterior chamber angle toward the ciliary processes. The innermost circular fibers, located the most anteriorly, form a ring parallel to the limbus.24 During accommodation, the fibers rearrange such that the circular portion thickens such that the innermost portion of the muscle moves even more inward and anterior.25

Although the ciliary muscle has been shown to retain its accommodative abilities throughout life26, a similar anterior/inward position to the “accommodative posture” is seen in aging ciliary muscles.26-27 This remodeling of the muscle has been proposed to be a mechanism to overcome the decrease in the index of refraction of the crystalline lens that occurs with age and which reduces the eye’s ability to accommodate effectively.27-28

These age-related changes in ciliary muscle structure were important reasons to exclude presbyopic patients from the current study.

The muscle fiber types found in the ciliary muscle vary according to their locations in each of the three bundles. The longitudinal fibers contain fewer mitochondria and more myofibrils than the radial and circular fibers, which make them resemble the fast fibers of striated muscle.24 It has been hypothesized that these fibers can thus contract faster than the rest of the ciliary muscle and supply the stiffness required for effective accommodation.24

Not completely separate from the question of how the myopic eye grows into a relatively more prolate shape, is why an eye might grow longer than the focal point formed by its refractive components. What stimulates this kind of growth? Certainly genetics plays a large role, as children of myopic parents are much more likely to be myopic23, but many researchers believe there is also an environmental role. Extrapolation of the well-established visual stimulus required for emmetropization29-30 has given rise to the most widely studied theory of myopic eye growth. This theory postulates that a hyperopic defocus signal on the retina stimulates the eye to elongate to meet that defocus.

The source and location of this defocus signal, however, has been debated over the years.

Originally, much emphasis was placed on the role of near work31-34 and accommodative lag17-19 in myopia development.

The association between myopia and the amount of near work a person does has been much studied. Angle & Wissmann (1980)31 used data from a survey of 12-17 year olds conducted from 1966-1970 to support what they termed this “use-abuse” theory of myopia development. The “use-abuse” theory postulates that an increase in myopia prevalence in more educated populations is a result of the greater amount of near work

7 done by more educated persons. The original theory was that people that spend more time focused on near tasks, such as reading or writing, would develop a refractive error that mirrored that posture, i.e., myopia. Richler & Bear (1980)32 found an association between refractive error and nearwork (hours per day) in a population of Newfoundlanders aged 5 and older, namely that myopic refractive error increased with an increase in hours of near work performed per day. In a study of 870 teenagers, Zylbermann et al. (1993)33 found that a subset of 193 Orthodox Jewish male students had a significantly higher prevalence and degree of myopia than the rest of the sample. They postulated that this may be due in part to their study habits that require sustained near vision and frequent changes in accommodation. Zadnik et al. (1994)34 and the OLSM (Orinda Longitudinal Study of

Myopia) study group found that even before the onset of myopia, children of myopic parents had longer eyes than children of non-myopic parents. This relationship persisted even when near work was controlled for, but the predictive model of myopia development that incorporated parental history was improved if near work was added to the model.

This association between myopia and near work naturally led to the thought that the amount and accuracy of accommodation done by an eye during its formative years may play a role in myopic eye growth. Studies of animal models show that if minus lenses are placed over a young animal’s eyes, the eyes grow longer to compensate for the imposed hyperopic defocus. When the minus lens is taken off, the eye’s refractive error emulates that of the initial defocusing lens. A similar result is seen for myopic defocus induced by plus lenses, although to a lesser extent in primates than is seen in chickens.35-

36 Anisometropia can be easily induced in monkeys if a plus lens is placed over one eye and a minus lens over the other.36

Although these experiments cannot ethically be performed in humans, studies of accommodative lag have offered insight into the role of defocus in human myopic eye growth. Gwiazda and colleagues have shown that myopic children show higher degrees of accommodative lag than emmetropic children for near targets. They postulated that this accommodative lag produces a hyperopic defocus, much like that seen in the animal lens compensation experiments, which could provide a stimulus for myopic eye growth.

17-18 Gwiazda and colleagues have also shown that myopic children have elevated AC/A

(accommodative convergence/accommodation) ratios, which may be the underlying cause of their increased accommodative lag. The child with a high AC/A ratio must relax accommodation to reduce accommodative convergence and maintain single binocular vision, which results in an accommodative lag.19 Whether these accommodative abnormalities precede the development of myopia is still debated. Gwiazda et al. showed that accommodation is reduced and AC/A ratios are elevated in myopic children prior to the onset of myopia.37 Mutti and colleagues of the CLEERE study group, however, have shown in a longitudinal study that this increase in accommodative lag in myopic children is not seen until after the onset of myopia.38 There is agreement, though, that the elevated accommodative lag seen in myopic children does not persist in adult subjects once myopia stops progressing.37, 39 Furthermore, unpublished data from this lab have shown that this association between myopic refractive error and higher degrees of accommodative lag in children is actually reversed in adults; accommodative lag is

9 significantly reduced in adult myopic subjects, i.e., they accommodate more accurately, as compared to their emmetropic counterparts. Thus, accommodative lag is likely not the cause of myopic eye development, although its role remains a relevant topic for further research.

Theories of near work and accommodative lag postulate that myopic eye growth develops as a result of defocus on the central retina, however, recent research points toward defocus on the peripheral retina as a stronger candidate for driving myopic eye growth.40 Smith et al. have shown that peripheral form deprivation (i.e., placing diffuser spectacles over the eyes that significantly reduce the amount of light reaching the retinas) can induce myopic eye growth in infant monkeys even when foveal vision is unrestricted.

Furthermore, recovery of these induced refractive errors occurred after removal of the diffuser spectacles even in the presence of a surgically ablated fovea.40 Despite this attention on defocus, however, some investigators maintain that myopic eye growth is simply a result of normal eye growth being equatorially restricted such that any remaining growth occurs predominantly in the axial direction, leading to a relatively more prolate ocular shape and thus a myopic refractive error.12 This study will further investigate the role of the ciliary muscle in accommodative lag.

The mechanism of accommodation has been well studied, but not extensively in the context of anisometropia. Most relevant to this study is the subject of accommodative lag in patients with anisometropia. Accommodative lag is the difference between the dioptric power of the accommodative stimulus and the measured accommodative response from the subject. If the target is set at 25 cm, or 4.00 D, and the subject only

10 accommodates 3.50 D, she is said to have an accommodative lag of +0.50 D. If, however, she over-accommodates to 4.50 D, she is said to have an accommodative lead of −0.50 D.

Accommodative lag is often diagrammed over a range of dioptric stimuli by an accommodative stimulus-response function. The 1:1 line represents the function over which accommodative response exactly matches accommodative stimulus, i.e., the subject accommodating perfectly accurately to each stimulus level with no lag or lead present. The non-linear portion of the curve (1) near 0.00 D stimulus is influenced primarily by tonic accommodative input and depth of focus, creating a lead of accommodation. The majority of the function is dominated by the linear portion of the function (2), which represents the accommodative response increasing at approximately the same rate as the accommodative stimulus. The slope of this portion is thus close to

1.0, yet it represents a steady state of accommodative lag, presumably because the accommodative system only needs to bring the object of regard into the range of depth of focus, rendering any additional accommodative effort useless. The upper limits of the curve (3), in which the slope levels off and eventually declines, represent the limit of accommodative amplitude for the subject. The subject can no longer produce enough accommodation to match the target stimulus and lag greatly increases.

Figure 1. Accommodative stimulus-response function (adapted from Borish41). Numbers represent the various zones of the function as described in the text.

In anisometropic subjects, accommodation is less well understood. Hering’s law of equal innervation led early investigators to assume that accommodation was equal between the two eyes during binocular viewing. Later studies however, challenged this assumption by showing that monocular accommodation between the two eyes of a subject may vary from 0.50 D42 to 0.75 D43 when the subject is viewing a target under binocular conditions. This indicated that if the eyes were anisometropic by ≤0.50-0.75 D, the eyes could compensate for this difference during accommodative tasks. In general, however, this inequality in accommodation averaged between 0.106 D and 0.213 D,

12 indicating that accommodation is relatively equal between the two eyes during binocular viewing.42, 44 In 1992, Flitcroft et al. showed that the accommodative response was a compromise between inputs to the two eyes when viewing an anisometropic stimulus.45

In the above studies, however, non-anisometropic subjects were used and anisometropia was simulated with lenses placed over one eye of the subject during binocular viewing.

Accommodation in naturally anisometropic subjects has not been well studied, but the prevailing assumption, validated in a study of marmosets, is that the accommodation of both eyes tends to match the response to the lower of the two demands during binocular viewing.46-47 In the present study, we will compare monocular accommodative lag between the two eyes of anisometropic subjects to investigate whether refractive error influences accommodative ability. We will test for any correlations between accommodative lag and ciliary muscle thickness between the two eyes of anisometropic subjects to further investigate the relationships between refractive error/axial length, ciliary muscle thickness, and accommodation within the context of anisometropia.

Aniso-accommodation has been proposed as a possible factor in myopia development. Charman suggested that the aniso-accommodative demands associated with near work may be the underlying cause of the association between near work and myopia. He showed that when reading, especially with a head tilt, the accommodative demands to the two eyes differ by up to 2.00 D by the end of a line of text. As it has been shown that the two eyes only have an ability to produce at most 0.50 D of aniso- accommodation, Charman reasoned that both eyes will experience a significant amount

13 of defocus during near work that is only exacerbated by accommodative lag.46 The implications for development of anisometropia, however, remain unclear.

The abundance of evidence cited above supports the associations of myopic refractive error with ciliary muscle thickness20-22, near work31-34, and accommodative lag17-19. As the ciliary muscle is clearly implicated in both near work and accommodative lag, further study of the ciliary muscle and its potential role in myopic eye growth is warranted. Any such study should aim to link both ciliary muscle structure and function, as both have been suggested by various researchers as playing a role in myopic eye growth. The function of the ciliary muscle in accommodation is implicated in the studies of near work31-34, and accommodative lag17-19 , while the physical structure as a potential source of equatorial growth restriction is suggested by the CLEERE study group.8 The present study aims to study ciliary muscle thickness (as a measure of physical structure) and accommodative lag (as a measure of muscle function) in a sample of patients with anisometropia of 1.00 D or greater. We will study the relationships between ciliary muscle thickness, accommodative lag, refractive error, and axial length in these subjects.

In addition, we will compare the two eyes within an anisometropic subject to determine if differences exist in ciliary muscle thickness and accommodative lag between the two eyes. The use of anisometropic subjects will elegantly eliminate the need to account for differences in genetic and environmental factors between individuals, as previous studies20-21 cannot refute the possibility that thicker ciliary muscles and longer, more myopic eyes are spuriously correlated due to the presence of background genetic or

14 environmental effects. Thus, we will be able to accurately assess if an increase in eye length alone results in an increased ciliary muscle thickness.

Methods

Subjects

Subjects were recruited via study advertisements including emails to faculty, staff, and students at The Ohio State University (OSU) College of Optometry, letters sent to patients of OSU Optometry Services who had a code of anisometropia in the computerized patient records, and also through flyers emailed to offices of local eye care practitioners and placed in the Optometry Services and the OSU Student Health Center.

Thirty subjects aged 18-40 years (mean = 28.08 years, SD = 5.5 years) with at least 1.00 D of spherical refractive difference between the two eyes (mean = 1.67 D, SD

= 1.15 D) were recruited to participate in the study. Exclusion criteria were amblyopia, mental disability that would prevent the subject from completing the testing protocol, or pregnancy by self-report. For the purposes of this study, amblyopia was defined as best- corrected visual acuity of worse than 20/40 in either eye as measured using a high contrast logMAR chart at 4 m under normal room illumination with habitual correction.

Written informed consent was obtained from each subject. The study was approved by the Institutional Review Board of The Ohio State University.

Refractive Error Measurements

All measurements were made on both eyes. Cycloplegia was achieved by instilling one drop of 0.5% proparacaine followed by two drops of 1% tropicamide spaced five minutes apart. Cycloplegic measurements were made 25 minutes after the second drop of tropicamide. Cycloplegic, spherical equivalent refractive error in each eye was obtained from the mean of ten readings with the Grand Seiko WV-500 (Grand Seiko

Co., Hiroshima, Japan) autorefractor. A Badal optometer system was used to place the fixation target of rows of letters at the most clear distance for the subject. The mean ± SD spherical equivalent refractive error was −2.46 D ± 3.27 (range −8.40 D, to +5.84 D).

Accommodative Measurements

Measurement of the accommodative response was performed monocularly using the Grand Seiko WV-500 autorefractor. All accommodative measurements were taken prior to the instillation of any cycloplegic eye drops. Subjects wore their habitual correction to look through a Badal optometer system at a target set first at 0.00 D and then at a 4.00 D stimulus level. The mean of ten readings at each stimulus level was used to calculate accommodative lag via the protocol described in the CLEERE study group investigations.48

In the event that a subject was uncorrected by greater than or equal to ±0.50 DS or

±1.00 DC in either eye (as determined by the average of the ten readings with the Grand

Seiko autorefractor) at the 0 D Badal stimulus through the subject’s habitual correction, a trial lens was placed over the subject’s eye and the measurements repeated. If the subject

17 was wearing contact lenses, the power of the trial lens reflected the amount of spherical over-refraction recommended by the autorefractor. If the subject was wearing glasses, the power of the trial lens reflected the combination of the over-refraction recommended by the autorefractor and the subject’s habitual spherical spectacle prescription as determined by neutralization of the glasses worn by the patient on the visit day.

Axial Length Measurements

Axial length (AL) measurements were made using the Zeiss IOL Master. Five consecutive measurements were taken on each eye and the average of these five measurements was used as the AL measurement for each eye. Only measurements with a signal to noise ratio of >2.0 were accepted to ensure only high confidence measurements were recorded.

Ciliary Muscle Thickness Measurements

Images of the nasal ciliary muscle of each eye were obtained with the Zeiss

Visante™ Anterior Segment OCT. Measurements were made under cycloplegia as previously described.21 Briefly, four images were obtained on the nasal ciliary muscle of each eye in the Enhanced High Resolution Corneal Mode. Radial ciliary muscle thickness measurements were made at the point of maximum ciliary muscle thickness (CMTMAX) and at 1.0 mm (CMT1), 2.0 mm (CMT2), and 3.0 mm (CMT3) posterior to the scleral spur using a semiautomatic extraction algorithm as described previously by Kao et al

(2011).49 Figure 2 is a representative image showing these measurements.

Analysis of the anatomy of the ciliary muscle shows that three distinct fiber types

(longitudinal, radial, and circular) are represented in differing proportions throughout the muscle. Thus it was reasonable to assume that the locations of our thickness measurements at 1.0 mm (CMT1), 2.0 mm (CMT2), and 3.0 mm (CMT3) behind the scleral spur, as well as the maximum thickness of the muscle (CMTMAX), were representative of different fiber types within the muscle. Based on existing histological studies of these fiber types24, we reasoned that the most posterior measurement (CMT3) likely represented only longitudinal fibers, the next most posterior measurement (CMT2) represented both longitudinal and some radial fibers, and the two most anterior measurements (CMT1 and CMTMAX) likely represented all three fiber types

(longitudinal, radial, and circular fibers). To parse out the portion of the muscle that represented mostly circular fibers, we subtracted the thickness of CMT2 (presumed longitudinal and radial fibers) from the thickness of CMT1 and CMTMAX (presumed combination of all fibers) to result in what we have termed “presumed radial/circular fibers atCMT1” and “presumed radial/circular fibers at CMTMAX.”

Statistical Analyses

Across-subject relationships between CMT and refractive error and axial length

While one of the goals for this study was to compare the shorter/more hyperopic eye to the longer/more myopic eye within a person, some of our analyses also sought to make comparisons across subjects. For these analyses, the mean refractive error and mean axial length for each subject was calculated in the following manner. Mean

19 refractive error was calculated by taking the mean of the refractive error for the right and left eyes of a subject, resulting in one mean refractive error for each subject. Mean axial length was calculated in the same way for each subject.

Each subject also had up to four CMT measurements taken for each location posterior to the scleral spur (CMTMAX, CMT1, CMT2, and CMT3) in both eyes. The mean of the four measurements was used to create one mean thickness measurement for each eye for each location posterior to the scleral spur. Simple linear regression models were used to plot CMT as a function of both mean spherical equivalent refractive error and mean axial length.

Across-subject relationship between the presumed circular/radial fibers at CMTMAX and

CMT1 and refractive error and axial length

Multilevel linear regression models were used to model the presumed radial/circular fibers at CMTMAX (CMTMAX – CMT2) and at CMT1 (CMT1 – CMT2) as functions of mean spherical equivalent refractive error and mean axial length. Age and gender were included in the model as covariates. The models had the form:

Outcomeij  A  B* SPHEQij ( or ALij )  C * Female  D Agei  ui   ij

In the model, i indexes the subject, and j her measure taken from eye j. The u term is a random effect that corrects the intercept (A) for CMT variation in subject i. It is needed to account for repeated within-subject measures of the outcome (one from each eye). For all models, age was centered at 28 years, mean spherical equivalent refractive error was centered at −3.00 D, and mean axial length was centered at 24 mm.

Within-subject relationship between CMT and refractive error and axial length

A similar multilevel regression model was calculated to model CMT against an indicator of the more myopic eye. Each CMT location was an outcome in its own model.

All models controlled for eye (right vs. left), mean spherical equivalent refractive error within a subject (as described above) centered at −3.00 D, gender, age centered at 25 years, and location of the width measurement (1.0 mm, 2.0 mm, or 3.0 mm posterior to the scleral spur). A comparable model controlling for mean axial length within a subject

(centered at 25 mm) instead of mean spherical equivalent refractive error was used for the comparisons of CMT with axial length. Continuous variables were centered to make model intercepts more meaningful (i.e. an outcome value for a 25-year-old subject).

Centering did not affect the slope estimates of the predictors.

Within-subject relationship between the presumed circular/radial fibers at CMTMAX and

CMT1 and refractive error and axial length

T-tests of differences between the more and less myopic eyes were performed for both presumed circular/radial fibers at CMTMAX and CMT1. Multilevel linear regression models were also fit for the two outcomes. The models had the form:

Outcomeij  A  B * avgSPHEQi ( or avgALi )  C * Female  D  Agei  E  RightEyeij 

F  MoreMyopicEyeij ( or LongerEyeij )  ui   ij i

In the model, i indexes the subject, and j his measure taken from eye j. The u term is a random effect that corrects the intercept (A) for CMT variation in subject i. It is needed 21 to account for repeated within-subject measures of the outcome (one from each eye). For all models, age was centered at 28 years, mean spherical equivalent refractive error was centered at −3.00 D, and mean axial length was centered at 24 mm.

Relationship between CMT and accommodative lag

T-tests of differences between the more and less myopic eyes were performed to assess within-subject differences in accommodative lag. A multilevel regression model was used to examine the across-subject relationships between accommodative lag, CMT, mean spherical equivalent refractive error and mean axial length. In this model, accommodative lag was the predictor, and the outcomes were CMT at all the various locations (1 mm, 2 mm, and 3 mm posterior to the scleral spur) as well as the maximum

CMT. Control variables in all models were age (centered at 28 years) and gender.

CMT3 CMT2 CMTMAX Scleral CMT1 Spur

Figure 2. Representative Visante™ image of the nasal ciliary body while the subject views an external fixation target. The thickness measurements 1 mm (CMT1), 2 mm (CMT2), and 3 mm (CMT3) posterior to the scleral spur are shown (pink). The maximum thickness of the ciliary body (CMTMAX) is also shown (yellow). 22

Results

General characteristics of the study sample are displayed in Table 1. Thirty subjects aged 21-40 years (mean = 28.08 years, SD = 5.5 years) with at least 1.00 D of spherical refractive difference between the two eyes (mean = 1.67 D, SD = 1.15 D) participated in the study. Mean spherical equivalent refractive error for the two eyes within each subject ranged from –8.40 D to +5.84 D (mean = –2.46 D, SD = 3.3 D), and mean axial length for the two eyes within each subject ranged from 21.36 mm to 26.51 mm (mean = 24.48 mm, SD = 1.4 mm). Ciliary muscle thickness was thickest at CMT1

(mean = 824.83 μm, SD = 76.3 μm), thinned at CMT2 (mean = 597.02 μm, SD = 99.4

μm), and was thinnest at CMT3 (mean = 348.67 μm, SD = 82.5 μm).

Table 1. Age, Spherical Equivalent Refractive Error, and Ciliary Muscle Thickness

Mean ± SD Median Min Max

Age (years) 28.08 ± 5.5 26.31 21.08 40.75

Refractive Error† (D) −2.46 ± 3.3 −3.19 −8.40 +5.84

Axial Length‡ (mm) 24.48 ± 1.4 24.69 21.36 26.51

Anisometropia (D) 1.67 ± 1.2 1.39 0.14 5.90

CMT1 (μm) 824.83 ± 76.3 818.81 586.25 963.75

CMT2 (μm) 597.02 ± 99.4 601.25 306.25 801.25

CMT3 (μm) 348.67 ± 82.5 343.13 156.25 475.00

†Mean refractive error for the two eyes within each subject ‡Mean axial length for the two eyes within each subject CMT: ciliary muscle thickness 1 mm (CMT1), 2 mm (CMT2) and 3 mm (CMT3) posterior to the scleral spur.

Across-subject relationship between CMT and refractive error and axial length

We recognized at the outset of this study that the cause of one eye becoming more myopic than the other within an anisometropic person may not be the same as that which causes a person to become myopic as a whole. Thus, we began by looking at relationships across subjects, rather than within a subject, to determine if the previously reported20-22 relationships between CMT and refractive error and axial length were present within this sample of anisometropes. The simple linear regressions of CMT measurements for each subject against the mean refractive error for each subject and mean axial length for each subject showed statistically significant correlations in the

24 negative direction for mean refractive error and in the positive direction for mean axial length (Figures 3 and 4). The strong negative correlation between CMT and mean refractive error meant that subjects with a more negative, or more myopic, mean refractive error tended to have thicker ciliary muscles than subjects with a more positive, or less myopic, mean refractive error. The positive correlation between CMT and mean axial length meant that subjects with longer eyes tended to have thicker ciliary muscles than subjects with shorter eyes. The greatest negative correlation was between CMT2 and mean refractive error (p < 0.0001), and the greatest positive correlation was between

CMT2 and mean axial length (p < 0.0001). After this evaluation of the data on a subject level, we were encouraged that our sample of anisometropes showed the same relationships between ciliary muscle thickness and both refractive error and axial length that the previous studies20-22 showed across individuals.

Figure 3. Mean ciliary muscle thickness plotted against mean spherical equivalent refractive error for each subject. For each subject, the mean spherical equivalent refractive error of the right and left eyes was calculated to give one mean spherical equivalent refractive error per subject. For each eye of each subject, the mean ciliary muscle thickness measurements at 1 mm (CMT 1, y = −11.21x + 797.3, R2 = 0.20, p = 0.0003), 2 mm (CMT 2, y = −22.01x + 542.8, R2 = 0.49, p > 0.0001) and 3 mm (CMT 3, y = −16.71x + 307.5, R2 = 0.40, p < 0.001) behind the scleral spur, as well as the maximum ciliary muscle thickness measurement (CMT MAX, y = −9.65x + 829.7, R2 = 0.14, p = 0.003), were then plotted against the mean spherical equivalent refractive error for that subject. Each subject thus had two data points for CMT1 (one for the right eye, one for the left eye) plotted against one mean spherical equivalent refractive error for each subject (the mean of the two eyes). This was repeated for CMT2, CMT3, and CMTMAX.

Figure 4. Mean ciliary muscle thickness plotted against mean axial length for each subject. For each subject, the mean of the axial length of the right and left eyes were calculated to give one mean axial length per subject. For each eye of each subject, the mean ciliary muscle thickness measurements at 1 mm (CMT 1, y = 23.97x + 238.2, R2 = 0.16, p = 0.002), 2 mm (CMT 2, y = 45.29x + −511.86, R2 = 0.35, p < 0.0001) and 3 mm (CMT 3, y = 34.55x + −497.3, R2 = 0.29, p < 0.0001) behind the scleral spur, as well as the maximum ciliary muscle thickness measurement (CMT MAX, y = 19.71x + 370.9, R2 = 0.10, p = 0.01), were then plotted against the mean axial length for that subject. Each subject thus had two data points for CMT1 (one for the right eye, one for the left eye) plotted against one mean axial length for each subject (the mean of the two eyes). This was repeated for CMT2, CMT3, and CMTMAX.

Across-subject relationship between the presumed circular/radial fibers at CMTMAX and

CMT1 and refractive error and axial length

Tables 2 and 3 show the multilevel linear regression model results for the presumed radial/circular fibers at CMT1 and CMTMAX. The data indicate that the thickness of the presumed circular/radial fibers at CMTMAX and CMT1 are correlated with mean refractive error and mean axial length in the opposite manner than CMT2 and

CMT3. In the model in Table 2, the presumed circular/radial fibers at CMT1 and

CMTMAX were significantly correlated with mean refractive error in a positive manner; i.e. subjects with a more hyperopic mean refractive error had thicker presumed circular/radial fibers at CMTMAX and CMT1, while the subjects with a more myopic mean refractive error tended to have thinner presumed circular/radial fibers CMTMAX and CMT1. A similar reversal of the trend found for CMT2 and CMT3 was noted for axial length in the models in Table 3. The negative correlation between axial length and the presumed circular fibers at CMT1 and CMTMAX suggests that subjects with shorter eyes on average had thicker presumed circular/radial fibers. Figures 5-8 show the relationships detailed in tables 2 and 3 in graphical form.

The data in tables 2 and 3 also show a significant positive association of age with the thickness of the presumed circular/radial fibers at CMTMAX and CMT1. This suggests that older subjects had thicker presumed circular/radial fibers than the younger subjects.

Table 2. The relationship between refractive error and the presumed circular/radial fibers at CMTMAX and CMT1

Predictor Presumed Circular/Radial Fibers CMTMAX CMT1 Intercept 263.0 235.3 11.65 10.34 Refractive Error† (p < 0.0001) (p < 0.0001) −17.0 −17.0 Female (p = 0.39 ) (p = 0.32 ) 3.07 2.82 Age (years) (p = 0.05) (p = 0.04) †Mean refractive error for the two eyes within each subject

Table 3. The relationship between axial length and the presumed circular/radial fibers at CMTMAX and CMT1

Predictor Presumed Circular/Radial Fibers CMTMAX CMT1 Intercept 277.9 247.5 −26.0 −22.4 Axial Length† (p = 0.0002) (p = 0.0002) −14.2 −13.8 Female (p = 0.51) (p = 0.47 ) 3.92 3.61 Age (years) (p = 0.02) (p = 0.02) †Mean axial length for the two eyes within each subject

Figure 5. Presumed circular fibers at CMT1 plotted against the mean spherical equivalent refractive error for each subject.

Figure 6. Presumed circular fibers at CMTMAX plotted against the mean spherical equivalent refractive error for each subject.

Figure 7. Presumed circular fibers at CMT1 plotted against the mean axial length for each subject.

Figure 8. Presumed circular fibers at CMTMAX plotted against the mean axial length for each subject.

Within-subject relationship between CMT and refractive error and axial length

In a multilevel regression analysis that compared the two eyes within these anisometropic subjects, ciliary muscle thickness of the more myopic eye was estimated to be the same as the less myopic eye for all locations of CMT (Table 4). The same relationship also held true for axial length; the multilevel regression analysis showed a significant positive correlation between ciliary muscle thickness and axial length, but within a subject, the ciliary muscle thickness of the longer eye was estimated to be the same as the shorter eye (Table 5). The data, however, did not even trend toward a thicker ciliary muscle in the more myopic or longer eye of the person. In fact, the trend was in the opposite direction, i.e., if there had been a statistically significant relationship, the more myopic and longer eye within a person would have had a thinner ciliary muscle than its fellow eye (Tables 4 and 5).

Table 4. Multilevel linear regression model for CMT and the more myopic eye within a person

Predictor CMTMAX CMT1 CMT2 CMT3 Intercept 797.5 760.3 459.0 242.8 Refractive −9.72 −11.3 −22.1 −16.8 Error† (p = 0.03) (p = 0.009) (p < 0.0001) (p = 0.0001) More Myopic −13.1 −12.0 −6.64 −2.81 Eye (p = 0.32) (p = 0.27) (p = 0.48) (p = 0.77) 12.76 12.19 9.14 −0.31 Right Eye (p = 0.33) (p = 0.26) (p = 0.34) (p = 0.97) 3.66 3.68 20.91 20.22 Female (p = 0.91) (p = 0.91) (p = 0.51) (p = 0.48) 1.81 1.57 −1.25 −0.66 Age (years) (p = 0.46) (p = 0.52) (p = 0.61) (p = 0.77) †Mean refractive error for the two eyes within each subject

Table 5. Multilevel linear regression model for CMT and the longer eye within a person

Predictor CMTMAX CMT1 CMT2 CMT3 Intercept 837.0 804.3 561.3 320.6 19.78 24.00 47.82 36.62 Axial Length† (p = 0. 07) (p = 0. 03) (p = 0. 0003) (p = 0. 001) −10.3 −9.78 −6.64 −2.50 Longer Eye (p = 0. 44) (p = 0. 37) (p = 0. 49) (p = 0. 80) 15.50 14.45 9.14 −0.00 Right Eye (p = 0. 25) (p = 0. 19) (p = 0. 35) (p = 1.00) 5.86 8.53 18.29 17.90 Female (p = 0. 87) (p = 0. 81) (p = 0. 64) (p = 0. 60) 1.43 1.23 −2.66 −1.76 Age (years) (p = 0. 60) (p = 0. 65) (p = 0. 37) (p = 0. 50) †Mean axial length for the two eyes within each subject

Within-subject relationship between the presumed circular/radial fibers at CMTMAX and

CMT1 and refractive error and axial length

The above analyses showed that the presumed circular fibers at CMTMAX and

CMT1 are positively correlated with mean refractive error of a subject and negatively correlated with mean axial length of a subject. Next, we applied multilevel linear regression models to examine the relationship between the presumed circular/radial fibers at CMTMAX and CMT1 and refractive error and axial length, comparing the two eyes of an anisometropic subject. Tables 6 and 7 show the results of this regression model. The presumed circular/fibers at CMTMAX and CMT1 were not thicker in the more myopic eye (Table 6) or the longer eye (Table 7) of an anisometropic person.

Table 6. Multilevel linear regression model comparing the presumed radial/circular fibers at CMTMAX and CMT1 between the two eyes of anisometropic subjects

Predictor Presumed Circular/Radial Fibers CMTMAX CMT1 Intercept 338.6 301.3 12.39 10.83 Refractive Error† (p <0.0001) (p <0.0001) More Myopic −6.44 −5.40 Eye (p = 0.65) (p = 0.66) 3.62 3.05 Right Eye (p = 0.80) (p = 0.80) −17.2 −17.2 Female (p = 0.38) (p = 0.32) 3.06 2.82 Age (years) (p = 0.05) (p = 0.04) †Mean refractive error for the two eyes within each subject

Table 7. Multilevel linear regression model comparing the presumed radial/circular fibers at CMTMAX and CMT1 between the two eyes of anisometropic subjects

Predictor Presumed Circular/Radial Fibers CMTMAX CMT1 Intercept 275.7 243.0 −28.0 −23.8 Axial Length† (p = 0.0003) (p = 0.0004) −3.70 −3.14 Longer Eye (p = 0.80) (p = 0.80) 6.36 5.31 Right Eye (p = 0.67) (p = 0.67) −12.4 −9.76 Female (p = 0.59) (p = 0.63) 4.09 3.89 Age (years) (p = 0.02) (p = 0.02) †Mean axial length for the two eyes within each subject

Relationship between CMT and accommodative lag

We next examined the relationship between ciliary muscle thickness and accommodative lag. A repeated measures regression model showed that accommodative lag to a 4.0-D target was significantly associated with CMT2 and with CMT3. The correlation is in the negative direction, indicating that a thicker posterior portion of the ciliary muscle is associated with a lower amount of accommodative lag.

Table 8. The relationship between accommodative lag for a 4.0-D stimulus and Ciliary Muscle Thickness

Predictor CMTMAX CMT1 CMT2 CMT3 Intercept 859.4 832.9 609.4 358.2 Accommodative −31.3 −39.4 −114.0 −104.0 Lag (4.0-D (p = 0. 43) (p = 0. 32) (p = 0. 02) (p = 0. 01) stimulus, D) 4.15 4.66 27.82 27.92 Female (p = 0. 91) (p = 0. 89) (p = 0. 51) (p = 0. 42) 3.26 3.40 4.19 4.38 Age (years) (p = 0. 33) (p = 0. 31) (p = 0. 30) (p = 0. 19)

Table 9 shows the results of paired t-tests to measure any difference in accommodative lag, CMT, refractive error, and axial length between the two eyes of an anisometropic subject in the current study sample. Not surprisingly, there is a significant difference in refractive error and axial length between the two eyes, i.e., these are patients with anisometropia, and, as reported above, there is no statistically significant difference between eyes in ciliary muscle thickness at any location tested. This result agrees with the regression models above that showed no difference in ciliary muscle thickness

35 between the two eyes of an anisometropic subject when the entire ciliary muscle is taken into account. Once the p-values are adjusted for multiple comparisons, the difference in accommodative lag between the two eyes of an anisometropic subject loses statistical significance. Accommodative lag, therefore, does not vary significantly between the two eyes of an anisometropic person.

Table 9. T-tests of the hypothesis that the mean difference between the two anisometropic eyes of each subject was zero for multiple parameters

95% Confidence

Variable Mean ± SD Interval p Value†

Δ Lag (4.0-D) 0.26 ± 0.63 (0.021 , 0.49) 0.03

Δ CMTMAX 12.22 ±70.83 (−14.23 , 38.67) 0.35

Δ CMT1 11.22 ± 58.21 (−10.5 , 33.0) 0.30

Δ CMT2 6.03 ± 51.17 (−13.08 , 25.13) 0.52

Δ CMT3 2.83 ± 51.02 (−16.22 , 21.88) 0.76

Δ Refractive Error 1.67 ± 1.15 (1.24 , 2.10) <.0001

Δ Axial Length −0.72 ± 0.53 (−0.92 , −0.52) <.0001

∆ = difference between eyes (more hyperopic eye – more myopic eye) †After adjusting for multiple comparisons (0.05/7 = 0.01), only refractive error and axial length are statistically significant (bold).

Discussion

Across-subject relationship between CMT and refractive error and axial length

This sample of anisometropic subjects behaves in accordance with the literature20-

22 in that subjects with longer, more myopic eyes (when taken as an average of the two eyes) tended to have thicker ciliary muscles. This was what we expected to find and confirms that this sample of anisometropic subjects do not follow a different trend from the rest of the population.

Across-subject relationship between presumed circular/radial fibers at CMTMAX and

CMT1 and refractive error and axial length

Here we found a surprising reversal of the trend noted above. When the presumed circular/radial fibers at CMTMAX and CMT1 are isolated from the rest of the muscle by factoring out the muscle thickness at CMT2, we found that subjects with longer, more myopic eyes (when taken as an average of the two eyes) tended to have thinner ciliary muscles. The significance of this is profound; it shows that although ciliary muscle thickness tends to increase with increasing myopic refractive error/increasing axial length, the trend reverses when the posterior, longitudinal fibers are factored out. This suggests that longer, more myopic eyes have thicker longitudinal fiber portions of their

38 ciliary muscles, while shorter, more hyperopic eyes have thicker circular/radial fiber portions of their ciliary muscles.

Oliveira, et al. first reported that increasing ciliary muscle thickness is associated with increasing axial myopia, but his group only examined the ciliary body at 2mm and

3mm posterior to the scleral spur20; they did not report any measurements as far anterior as 1mm posterior to the scleral spur. Muftuoglu, et al. claim that their measurements of ciliary body and muscle thickness were taken at the thickest portion of the ciliary body22, which would theoretically correspond to our CMTMAX measurements; however, without any anchor from which to measure, such as the scleral spur, it may be impossible to tell exactly where on the ciliary body their measurements were taken. Furthermore, their ciliary body measurements took the ciliary processes into account as well, so the thickest portion of the ciliary body may not have been the thickest portion of the ciliary muscle. Thus, our lab is the first to definitively study the ciliary muscle at such an anterior position as 1 mm posterior to the scleral spur in vivo.

Oliveira et al. also found that CMT3 was positively associated with axial length, but did not find an association with refractive error when both axial length and refractive error were included in the model. They postulated that this was because the major association of CMT was with axial length, with refractive error being an essentially spurious correlation.20 The present study, however, finds that both refractive error and axial length are strong predictors of CMT in a multivariate analysis, and that the longer eye within a patient did not trend towards having a thicker ciliary muscle. This difference between the two studies can be accounted for by the difference in mean age of the

39 subjects in the present study as compared to Oliveira et al. The subjects in Oliveira’s study had a mean age of 51.8 ± 16.5 years20, while this study’s subjects had a mean age of 28.08 ± 5.5 years to purposely exclude the effects of disuse during presbyopia on

CMT. Therefore, it is likely that the effect of axial length on CMT in their sample was stronger than refractive error because the subjects were older. The refractive error of the subjects in that study was likely not as correlated with their axial length due to lenticular changes; either hyperopic shifts because of a more homogenous refractive index of the crystalline lens28, 50 or myopic shifts due to early nuclear sclerotic cataracts.50 Thus, their refractive error no longer matched their axial length like it did when they were younger.

The finding that the presumed circular/radial fibers at CMTMAX and CMT1 are significantly positively associated with age is consistent with the evidence that the longitudinal and radial portions of the ciliary muscle decrease in volume with age, while the circular portion increases with age.27

Within-subject relationship between CMT and refractive error and axial length

These results showed that the ciliary muscle thickness of the more myopic eye was estimated to be the same as the less myopic eye and the longer eye within an anisometropic subject. This finding was not expected based on the previous studies.20-22

At the outset of this study, we expected to find that the more myopic and longer eye within an anisometropic person would have a thicker ciliary muscle. Not only did we not find any statistically significant difference in CMT between the two eyes of a subject, the trend was not even in the expected direction. The results show an insignificantly thinner

CMT in the longer, more myopic eye. This indicated that the failure to find a significant difference in ciliary muscle thickness between the two eyes of an anisometropic person probably should not be attributed to an underpowered study. Rather, the data show that an eye can grow to become more myopic than its fellow eye without resulting in an increase in ciliary muscle thickness. This seemingly does not fit with the aforementioned cross- sectional studies,20-22 but data from a longitudinal study of ciliary muscle thickness in children performed in our lab confirm that the ciliary muscle does not necessarily thicken with eye growth.51

This finding is in direct contrast with that of Muftuoglu et al., who found an increase in CMT in the more myopic and longer eye of patients with unilateral high myopia.22 It is important to note, however, that the subjects in that study had a level of anisometropia not normally encountered in the general population. The prevalence of that degree of anisometropia (≥ 5.00 D) is unknown, but one study estimates that the prevalence of anisometropia ≥ 3.00 D in a German/Austrian population was only 1.4%.6

It is possible that the mechanisms regulating eye growth in such a high degree of anisometropia are different from the mechanisms driving eye growth in lower amounts of anisometropia, such as the sample in this study.

Within-subject relationship between presumed circular/radial fibers at CMTMAX and

CMT1 and refractive error and axial length

Here again we found that the more myopic and longer eye within an anisometropic person did not have a thickened presumed circular/radial fibers at

CMTMAX and CMT1. Again, this contrasts with the results found by Muftuoglu et al., who measured the ciliary muscle at the thickest part of the ciliary body during an ultrasound biomicroscope scan.22 Presumably this corresponds to the CMTMAX measurements in our study, but a direct comparison between the two studies is difficult because of the difference in measurement technique. Muftuoglu et al. used ultrasound biomicroscopy to scan the ciliary body parallel to the limbus to find the thickest portion.

The images obtained are in the opposite orientation to those obtained by the Visante anterior segment OCT in our study, thus it is impossible to compare the thickness of the muscle at that point to the thickness of the muscle along the entire antero-posterior axis of the muscle to ensure that indeed the thickest portion is being measured. Furthermore, the ultrasound biomicroscopy images in Muftuoglu’s study capture both the ciliary muscle and the ciliary processes. The ciliary muscle measurements were made by simply excluding the ciliary processes from the caliper measurements, as opposed to the imaging algorithm employed by our study that decidedly captures only the ciliary muscle.49 The

Visante anterior segment OCT is incapable of imaging the ciliary processes due to the density of pigment in the outer ciliary body epithelium.

Relationship between CMT and accommodative lag

In the present study, when we compared across individuals, we found that accommodative lag to a 4.0-D target was significantly negatively associated with CMT2 and CMT3, indicating that subjects with a thicker posterior portion of their ciliary muscles tended to have lower levels of accommodative lag. This trend is also seen in the

42 unpublished data from two other studies undertaken by this lab; a larger-scale study on a sample of 92 adults that were not specifically recruited for anisometropia52, and an even larger-scale study on a sample of 97 children.51 These two studies only tested the right eyes of each subject and the subjects were not screened for anisometropia.

When we compared within individual subjects, we found that accommodative lag is not statistically different between the two eyes of an anisometropic subject. This is not surprising given that we also found no difference in ciliary muscle thickness between the two eyes of an anisometropic subject. This indicates that structure and function of the ciliary muscle are indeed linked, as it can be assumed that the two eyes within an anisometropic person have been “reared” under identical stimuli to accommodation.

Implications

Our study provided some interesting insights into the possible mechanisms underlying myopic eye growth. First, it showed that although anisometropic subjects minded the previously observed association between thicker ciliary muscles and increasing myopia and axial length when compared across subjects, that same relationship was not observed when the two eyes within a subject were compared. In other words, if a subject has more myopic and longer eyes on average, she tended to have thicker ciliary muscles in her eyes when compared to someone less myopic than her, however, the more myopic and longer of her two eyes did not have a statistically significantly thicker ciliary muscle than the fellow eye. This indicated that an eye can grow longer and increase in myopic refractive error without necessitating an increase in

43 ciliary muscle thickness. Thus, the theory that the ciliary muscle somehow acts as a physically restrictive force to exacerbate eye growth in the axial direction, as has been suggested,8 seems unlikely given the data presented in this study and in a study of children from our laboratory that did not find a relationship between increases in axial length and the ciliary muscle.51 Rather, these data suggest that universal genetic and/or environmental mechanisms may govern both ciliary muscle thickness and eye growth such that a person who develops longer, more myopic eyes on average will also develop thicker ciliary muscles in those eyes. At a minimum, the data suggest the eye can elongate in myopia without an equatorial restriction to growth.

Second, this study showed that across subjects, longer and more myopic eyes tended to have thicker, more developed posterior portions of their ciliary muscles, while shorter and more hyperopic eyes tended to have thicker, more developed anterior portions of their ciliary muscles. The anterior and posterior portions of the muscle for this analysis are delineated at CMT2. The relationship between CMT and accommodative lag also showed a difference in the anterior and posterior portions of the muscle when split at the

CMT2 location. Only a thicker posterior portion of the ciliary muscle was associated with a lower amount of accommodative lag. Taken together with the unpublished results of the

92 non-anisometropic adult subjects from this lab, which showed that more myopic subjects tended to have lesser amounts of lag, these results suggest that more myopic subjects possess thicker posterior portions of their ciliary muscles and accommodate more accurately than their more emmetropic or hyperopic counterparts. On the other hand, the less myopic, more hyperopic subjects had thicker, more developed anterior

44 portions of their ciliary muscle, yet accommodated less accurately. The implications of these findings are that the anterior portion of the ciliary muscle, representing the presumed circular/radial fibers, may be thicker in more hyperopic eyes as a result of increased accommodative workload, and that the posterior, presumably longitudinal, fibers that are thicker in more myopic eyes play a large role in sustaining accurate accommodation. These findings appear to be consistent with what is known of ciliary muscle anatomy.

As discussed above, the various fiber orientations within the ciliary muscle also contain different physiological characteristics. The longitudinal fibers contain fewer mitochondria than the radial and circular fibers, thus resembling more the fast fibers of striated muscle. It has been suggested previously that these posterior, longitudinal fibers can contract faster than the rest of the ciliary muscle and thus offer the stiffness required for effective accommodation.24 Data from this study support this notion, as it is likely that this “stiffness” contributes to sustaining accommodation and reducing the amount of accommodative lag.

The separation between the anterior and posterior fibers of the ciliary muscle in both structure and function is echoed in the findings of another study from our laboratory that postulated a “fulcrum” point located approximately at the location of CMT2. The ciliary muscle generally thickened during accommodation anterior to this point and thinned during accommodation posterior to this point.53 Evidence of a “fulcrum” point regarding the structure of the zonular apparatus and thus influencing the mechanism of accommodation has also been described.25 Studies of aging ciliary muscles in humans

45 also show a differential in morphological changes between the anterior circular fibers, whose area increases with age, and the posterior longitudinal fibers, whose area decreases with age.27 This association between increased age and thicker presumed circular/radial fibers at CMTMAX/CMT1 was also observed in the present study. If the anterior circular fibers are truly associated with accommodative workload, then their thickening with age could conceivably represent the futile efforts of the ciliary muscle to effect accommodative lens changes in an eye progressing towards and into presbyopia.

The present study is somewhat limited by the anterior segment OCT imaging technique used to measure ciliary muscle thickness. It is not possible with this technique to ensure that each image is captured at exactly the same image plane because there is no easily identifiable landmark in the ciliary muscle to guide alignment. The newest version of Visante software (Version 3.0), however, may help with this alignment by providing a scanning beam on the surface of the globe much like the corneal reflex that guides alignment of corneal and crystalline lens images with the Visante. In the present study, within-subject variability was mostly eliminated by taking multiple images and using the mean of multiple measurements from those images in analysis.

The use of anisometropic subjects, while one of the strengths of the study, is also a limitation in that we cannot rule out the possibility that whatever caused one eye to grow longer than the other within a single subject is not, in fact, the same mechanism of myopic eye growth seen in the population at large. We also made some overriding assumptions with regards to the composition of ciliary muscle fiber types at differing

46 locations in the muscle. To truly verify these assumptions, histological comparison would be ideal.

Another limitation to the present study is the cross-sectional nature of the study.

Anisometropic subjects were used to simulate the effect of actual temporal eye growth, with the reasoning that there was clearly a difference in eye growth at some point between the two eyes that did not occur from the universal genetic and environmental background that was present for both eyes. While this provided useful clues to myopic eye growth, there is still no substitute for longitudinal studies that measure ciliary muscle thickness and accommodative lag before, during, and after the period of myopic refractive error development. Cross-sectional studies such as this one can only provide correlations; they cannot define a causal relationship between any parameters. Thus, longitudinal studies in children are still needed to further investigate the role that accommodation may play in ciliary muscle development and how accommodation and the ciliary muscle may be related to myopic eye growth.

References

1. Vitale S, Cotch MF, Sperduto R, Ellwein L. Costs of refractive correction of distance vision impairment in the United States, 1999-2002. Ophthalmology. Dec 2006;113(12):2163-2170. 2. Sperduto RD, Seigel D, Roberts J, Rowland M. Prevalence of myopia in the United States. Arch Ophthalmol. Mar 1983;101(3):405-407. 3. Vitale S, Ellwein L, Cotch MF, Ferris FL, 3rd, Sperduto R. Prevalence of refractive error in the United States, 1999-2004. Arch Ophthalmol. Aug 2008;126(8):1111-1119. 4. Vitale S, Sperduto RD, Ferris FL, 3rd. Increased prevalence of myopia in the United States between 1971-1972 and 1999-2004. Arch Ophthalmol. Dec 2009;127(12):1632-1639. 5. van AG. On emmetropia and ametropia. Opt Acta (Lond). 1961;142(Suppl):1-92. 6. Linke SJ, Richard G, Katz T. Prevalence and associations of anisometropia with spherical ametropia, cylindrical power, age, and sex in refractive surgery candidates. Invest Ophthalmol Vis Sci. Sep 2011;52(10):7538-7547. 7. Hashemi H, Khabazkhoob M, Yekta A, Mohammad K, Fotouhi A. Prevalence and risk factors for anisometropia in the Tehran eye study, Iran. Ophthalmic Epidemiol. Jun 2011;18(3):122-128. 8. Mutti DO, Sholtz RI, Friedman NE, Zadnik K. Peripheral refraction and ocular shape in children. Invest Ophthalmol Vis Sci. Apr 2000;41(5):1022-1030. 9. Atchison DA, Jones CE, Schmid KL, et al. Eye shape in emmetropia and myopia. Invest Ophthalmol Vis Sci. Oct 2004;45(10):3380-3386. 10. Logan NS, Gilmartin B, Wildsoet CF, Dunne MC. Posterior retinal contour in adult human anisomyopia. Invest Ophthalmol Vis Sci. Jul 2004;45(7):2152-2162. 11. Manns F, Parel JM, Denham D, et al. Optomechanical response of human and monkey lenses in a lens stretcher. Invest Ophthalmol Vis Sci. Jul 2007;48(7):3260-3268. 12. Mutti DO, Mitchell GL, Sinnott LT, et al. Corneal and Crystalline Lens Dimensions Before and After Myopia Onset. Optom Vis Sci. Mar 2012;89(3):251- 262. 13. Mutti DO, Hayes JR, Mitchell GL, et al. Refractive error, axial length, and relative peripheral refractive error before and after the onset of myopia. Invest Ophthalmol Vis Sci. Jun 2007;48(6):2510-2519. 14. Lossing LA, Sinnott LT, Kao CY, Richdale K, Bailey MD. Measuring Changes in Ciliary Muscle Thickness with Accommodation in Young Adults. Optom Vis Sci. Apr 12 2012. 48

15. Beers AP, Van Der Heijde GL. In vivo determination of the biomechanical properties of the component elements of the accommodation mechanism. Vision Res. Nov 1994;34(21):2897-2905. 16. Helmholtz Hv, Southall JPC. Helmholtz's treatise on physiological optics. Rochester, N.Y.: The Optical Society of America; 1924. 17. Gwiazda J, Thorn F, Bauer J, Held R. Myopic children show insufficient accommodative response to blur. Invest Ophthalmol Vis Sci. Mar 1993;34(3):690- 694. 18. Gwiazda J, Bauer J, Thorn F, Held R. A dynamic relationship between myopia and blur-driven accommodation in school-aged children. Vision Res. May 1995;35(9):1299-1304. 19. Gwiazda J, Grice K, Thorn F. Response AC/A ratios are elevated in myopic children. Ophthalmic Physiol Opt. Mar 1999;19(2):173-179. 20. Oliveira C, Tello C, Liebmann JM, Ritch R. Ciliary body thickness increases with increasing axial myopia. Am J Ophthalmol. Aug 2005;140(2):324-325. 21. Bailey MD, Sinnott LT, Mutti DO. Ciliary body thickness and refractive error in children. Invest Ophthalmol Vis Sci. Oct 2008;49(10):4353-4360. 22. Muftuoglu O, Hosal BM, Zilelioglu G. Ciliary body thickness in unilateral high axial myopia. Eye. May 2009;23(5):1176-1181. 23. Mutti DO, Mitchell GL, Moeschberger ML, Jones LA, Zadnik K. Parental myopia, near work, school achievement, and children's refractive error. Invest Ophthalmol Vis Sci. Dec 2002;43(12):3633-3640. 24. Tamm ER, Lutjen-Drecoll E. Ciliary body. Microsc Res Tech. Apr 1 1996;33(5):390-439. 25. Rohen JW. Scanning electron microscopic studies of the zonular apparatus in human and monkey eyes. Invest Ophthalmol Vis Sci. Feb 1979;18(2):133-144. 26. Strenk SA, Strenk LM, Guo S. Magnetic resonance imaging of aging, accommodating, phakic, and pseudophakic ciliary muscle diameters. J Cataract Refract Surg. Nov 2006;32(11):1792-1798. 27. Tamm S, Tamm E, Rohen JW. Age-related changes of the human ciliary muscle. A quantitative morphometric study. Mech Ageing Dev. Feb 1992;62(2):209-221. 28. Koretz JF, Handelman GH, Brown NP. Analysis of human crystalline lens curvature as a function of accommodative state and age. Vision Res. 1984;24(10):1141-1151. 29. Wildsoet CF. Active emmetropization--evidence for its existence and ramifications for clinical practice. Ophthalmic Physiol Opt. Jul 1997;17(4):279- 290. 30. Mutti DO, Mitchell GL, Jones LA, et al. Axial growth and changes in lenticular and corneal power during emmetropization in infants. Invest Ophthalmol Vis Sci. Sep 2005;46(9):3074-3080. 31. Angle J, Wissmann DA. The epidemiology of myopia. Am J Epidemiol. Feb 1980;111(2):220-228. 32. Richler A, Bear JC. Refraction, nearwork and education. A population study in Newfoundland. Acta Ophthalmol (Copenh). Jun 1980;58(3):468-478. 49

33. Zylbermann R, Landau D, Berson D. The influence of study habits on myopia in Jewish teenagers. J Pediatr Ophthalmol Strabismus. Sep-Oct 1993;30(5):319-322. 34. Zadnik K, Satariano WA, Mutti DO, Sholtz RI, Adams AJ. The effect of parental history of myopia on children's eye size. JAMA. May 4 1994;271(17):1323-1327. 35. Schaeffel F, Glasser A, Howland HC. Accommodation, refractive error and eye growth in chickens. Vision Res. 1988;28(5):639-657. 36. Smith EL, 3rd, Hung LF. The role of optical defocus in regulating refractive development in infant monkeys. Vision Res. Apr 1999;39(8):1415-1435. 37. Gwiazda J, Thorn F, Held R. Accommodation, accommodative convergence, and response AC/A ratios before and at the onset of myopia in children. Optom Vis Sci. Apr 2005;82(4):273-278. 38. Mutti DO, Mitchell GL, Hayes JR, et al. Accommodative lag before and after the onset of myopia. Invest Ophthalmol Vis Sci. Mar 2006;47(3):837-846. 39. Abbott ML, Schmid KL, Strang NC. Differences in the accommodation stimulus response curves of adult myopes and emmetropes. Ophthalmic Physiol Opt. Jan 1998;18(1):13-20. 40. Smith EL, 3rd, Kee CS, Ramamirtham R, Qiao-Grider Y, Hung LF. Peripheral vision can influence eye growth and refractive development in infant monkeys. Invest Ophthalmol Vis Sci. Nov 2005;46(11):3965-3972. 41. Benjamin WJ, Borish IM. Borish's clinical refraction. 2nd ed. St. Louis Mo.: Butterworth Heinemann/Elsevier; 2006. 42. Stoddard KB, Morgan MW. Monocular accommodation. American Journal of Optometry and Archives of American Academy of Optometry. 1942(November):460-465. 43. Marran L, Schor CM. Lens induced aniso-accommodation. Vision Res. Nov 1998;38(22):3601-3619. 44. Ball EA. A study of consensual accommodation. Am J Optom Arch Am Acad Optom. Nov 1952;29(11):561-574. 45. Flitcroft DI, Judge SJ, Morley JW. Binocular interactions in accommodation control: effects of anisometropic stimuli. J Neurosci. Jan 1992;12(1):188-203. 46. Charman WN. Aniso-accommodation as a possible factor in myopia development. Ophthalmic Physiol Opt. Sep 2004;24(5):471-479. 47. Troilo D, Quinn N, Baker K. Accommodation and induced myopia in marmosets. Vision Res. Apr 2007;47(9):1228-1244. 48. Mutti DO, Jones LA, Moeschberger ML, Zadnik K. AC/A ratio, age, and refractive error in children. Invest Ophthalmol Vis Sci. Aug 2000;41(9):2469- 2478. 49. Kao CY, Richdale K, Sinnott LT, Grillott LE, Bailey MD. Semiautomatic extraction algorithm for images of the ciliary muscle. Optom Vis Sci. Feb 2011;88(2):275-289. 50. Grosvenor T, Skeates PD. Is there a hyperopic shift in myopic eyes during the presbyopic years? Clin Exp Optom. Nov-Dec 1999;82(6):236-243.

51. Bailey MD, Sinnott LT, Kao CY. Longitudinal Ciliary Muscle Growth Is Not Correlated With Overall Eye Growth. 2010 Annual Meeting Abstract and Program Planner accessed at www.arvo.org. 2010:Abstract 2838. 52. Ernst LE, Sinnott LT, Bailey MD. Ciliary body thickness, refractive error, and axial length in adults. 2008 Annual Meeting Abstract and Program Planner accessed at www.arvo.org. 2008:Abstract 3580. 53. Lewis HA, Sinnott LT, Kao CY, Bailey MD. Changes in ciliary muscle thickness during accommodation in children. Optom Vis Sci. 2012;Accepted 02/08/2012.