Quantitative Analysis of the Structure of the Human
Extraocular Muscle Pulley System
Reika Kono,1,2 Vadims Poukens,1 and Joseph L. Demer1,3
PURPOSE. Extraocular muscle (EOM) paths are constrained by
connective tissue pulleys serving as functional origins. The
quantitative structural features of pulleys and their intercouplings and orbital suspensions remain undetermined. This
study was designed to quantify the composition of EOM pulleys and suspensory tissues.
METHODS. Five human orbits, ages 33 weeks gestation to 93
years, were imaged intact by magnetic resonance (MRI), serially sectioned at 10 ␮m thickness, and stained for collagen,
elastin, and smooth muscle (SM). With MRI used as a reference,
digital images of sections were geometrically corrected for
shrinkage and processing deformations, and normalized to
standard normal adult globe diameter. EOM pulleys, interconnections, suspensory tissues, and entheses were quantitatively
analyzed for collagen, elastin, and SM thickness and density.
RESULTS. Rectus and inferior oblique pulleys had uniform structural features in all specimens, comprising a dense EOM encirclement by collagen 1 to 2 mm thick. Elastin distribution
varied, but was greatest in the orbital suspension of the medial
rectus pulley and in a band from it to the inferior rectus pulley.
This region corresponded to maximum SM density. Structural
features of pulleys, intercouplings, and entheses were similar
among specimens. The major mechanical couplings to the
osseous orbit were near the medial and lateral rectus pulleys.
CONCLUSIONS. Quantitative analysis of structure and composition of EOM pulleys and their suspensions is consistent with in
vivo MRI observations showing discrete inflections in EOM
paths that shift predictably with gaze. Focal SM distributions in
the suspensions suggest distinct roles in stiffening as well as
shifting rectus pulleys. (Invest Ophthalmol Vis Sci. 2002;43:
2923–2932)
From the Departments of 1Ophthalmology and 3Neurology, University of California, Los Angeles; and 2Department of Ophthalmology,
Okayama University Medical School, Okayama, Japan.
Supported by National Eye Institute Grant EY08313 (JLD). JLD is
the recipient of an unrestricted award from Research to Prevent Blindness and is the Laraine and David Gerber Professor of Ophthalmology.
Submitted for publication December 19, 2001; revised May 1,
2002; accepted May 23, 2002.
Commercial relationships policy: N.
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Corresponding author: Joseph L. Demer, Jules Stein Eye Institute,
100 Stein Plaza, UCLA, Los Angeles, CA 90095-7002; jld@ucla.edu.
of connective tissue septa within the orbit and stereotypic
organization of connective tissue around the EOMs.
Modern interest in orbital connective tissues was reawakened by the demands of biomechanical modeling of binocular
alignment. Failures of initial attempts to compute ocular position based on static force equilibrium10 clarified the critical
importance of EOM path determinants that had to be assumed
arbitrarily in initial binocular models.11 The need for EOM path
data motivated early radiographic studies in monkeys12 and
humans,13 showing absence of obvious EOM sideslip over the
globe during ocular rotations. A decade ago Miller14 used
relatively low-resolution MRI with three-dimensional reconstruction to demonstrate the stability of rectus EOM belly paths
throughout the oculomotor range. The further demonstration
by MRI that EOM paths are little affected by large surgical
transpositions of their tendons provided strong evidence for
EOM path constraint by pulleys coupled to the orbit.15,16
Recent histologic studies have demonstrated that each rectus pulley consists of an encircling ring of collagen located near
the globe equator in Tenon fascia,17 coupled to the orbital
wall, adjacent EOMs, and equatorial Tenon fascia by slinglike
bands containing densely woven collagen,18 elastin, and
SM.17,19 Pulleys inflect rectus and inferior oblique (IO) EOM
paths in the same manner that the trochlea inflects the path of
the superior oblique (SO) tendon. The coronal plane location
of each rectus pulley has been shown by MRI to be highly
uniform in normal subjects.20 The global layer of each rectus
EOM, containing about half of all EOM fibers,21 passes through
the pulley and becomes contiguous with tendon to insert on
the globe. The orbital layer, containing the remaining half of
the EOM fibers, inserts on the pulley, not on the globe.21,22
The mechanical properties of pulleys are critical for ocular
kinematics, the rotational properties of the eye. Rotations of
any three-dimensional object are not mathematically commutative—that is, final eye orientation depends on the order of
rotations.23 This vexing conundrum regarding the neural control of ocular motility is avoided if the ocular rotational axis
shifts by half of the change in ocular orientation in relation to
a primary position, for under these conditions the effect of
noncommutativity becomes negligible.24 This half angle behavior is equivalent to Listing’s law of ocular torsion25 and is
faithfully observed when the head is upright and stationary.26
Pulleys appear important to Listing’s law and commutativity.
Precise mechanical shifts in rectus pulley positions are consistent with commutativity22 and have been quantitatively confirmed in humans by MRI in tertiary gaze positions.27 The
coordinated control postulate of the active pulley hypothesis
states that, in an oculocentric coordinate system, rectus pulley
location is maintained in a constant relationship with the
EOM’s scleral insertion, so that the distance from the pulley to
globe center is equal to the distance from globe center to
insertion. By this relationship, the velocity vector produced by
contraction of the EOMs shifts by half of the change in ocular
orientation from primary position. The IO muscle also has a
pulley, mechanically coupled to the inferior rectus (IR) pulley,28 that moves anteroposteriorly by half the travel of the IR
pulley to maintain an IR action perpendicular to the half-angle
behavior of the rectus EOMs.29,30 During convergence, rectus
Investigative Ophthalmology & Visual Science, September 2002, Vol. 43, No. 9
Copyright © Association for Research in Vision and Ophthalmology
2923
L
argely neglected because of the appeal of the impressive
extraocular muscles (EOMs), the neighboring orbital connective tissues have aroused only sporadic scientific interest.
Still, from the time of Lockwood,1 investigators have described
networks of connective tissues in the orbit and even the occurrence of smooth muscle (SM) cells.2– 4 The idea that these
connective tissues might function as pulleys for EOMs received
brief consideration5 before becoming largely forgotten. More
recently, Koornneef 4,6 –9 performed systematic histologic studies of the orbital tissues, describing the stereotypic occurrence
2924
Kono et al.
pulleys may systematically shift in the coronal plane to meet
visual demands, probably under the influence of the orbital
layers of the oblique EOMs31,32 and SM in the medial orbit.19
These observations indicate a far greater degree of complexity in EOM function than previously suspected, with mutual
mechanical interactions among EOMs in the periphery replacing the notion of a neural final common pathway.33 Interpretation of neural control of binocular coordination will require
an accurate computational simulation of the mechanics of the
EOMs and associated connective tissues. Although existing
computational models of orbital biomechanics now incorporate pulleys,34,35 these are nonphysiologic regarding their essentially fixed orbital suspensions and absence of realistic gazerelated movements. Physiologic models of EOM action must
incorporate a quantitative description of pulley structure. The
present study was performed to determine quantitatively the
composition and relative abundance of connective tissue components in the human EOM pulley system.
METHODS
En Bloc Tissue Preparation
All human specimens were obtained from cadavers in conformity with
legal requirements and in compliance with the tenets of the Declaration of Helsinki. Three human orbits (17 months, 4 years, and 57 years)
were obtained from a tissue bank (IIAM, Scranton, PA), in heads fresh
frozen shortly after death. The frozen heads were slowly thawed in
10% neutral buffered formalin. The head of a 33-week stillborn fetus
was obtained at autopsy and fixed by immersion in formalin. These
orbits were fixed in situ within the cranial bones. One additional
human orbit (93 years old) was exenterated en bloc at autopsy through
an intracranial approach within 24 hours of death and fixed for at least
5 days in 10% neutral buffered formalin with the periorbita intact but
separated from bony support.
Magnetic Resonance Imaging
The four fixed heads and the 93-year-old orbit were imaged by MRI,
with dual 3-in. phased array surface coils in a 1.5-T scanner (Signa;
General Electric, Milwaukee, WI). Multiple contiguous quasicoronal
MRIs, 2 or 3 mm in thickness, were then obtained with a T1 pulse
sequence encompassing a 256 ⫻ 256 matrix over a 4- or 5-cm2 field of
view, providing a pixel resolution of 156 or 195 ␮m, respectively. To
improve the signal-to-noise ratio at these high resolutions, four to nine
excitations were performed. Digital MRIs were transferred to computers (Macintosh; Apple Computer, Cupertino, CA), converted into eightbit tagged image file format (TIFF) by locally developed software, and
quantified by NIH Image (W. Rasband, National Institutes of Health;
available by file transfer protocol from zippy.nimh.nih.gov or on floppy
disc from NTIS Springfield, VA; part number PB95500195GEI).
Histologic Processing
The four orbits fixed in situ were then removed in continuity with the
eyelids and orbital bones. The orbital rims and walls were carefully
thinned under magnification by using a high-speed drill and rongeurs
before decalcification for 24 hours at room temperature in 0.003 M
EDTA and 1.35 N HCl.22
Formalin-fixed tissues were dehydrated in graded solutions of alcohol and chloroform or xylenes, embedded in paraffin, and serially
sectioned in the coronal plane at 10-␮m thickness by disposable metal
blades on a microtome (HM325; Carl Zeiss, Thornwood, NY), as previously described.17,19,22 This produced 2800 to 4800 sections per
orbit, depending on the orbit’s size. Alternate sets of five contiguous
sections were saved and mounted on 50 ⫻ 75 mm gelatin-coated glass
slides. Masson trichrome stain was used to show muscle and collagen,
and van Gieson stain to show elastin.36 As previously described,19 SM
was confirmed using a monoclonal mouse antibody to human SM
IOVS, September 2002, Vol. 43, No. 9
␣-actin, which was visualized with an avidin-biotin complex (ABC) kit
with blue chromogen (Alkaline Phosphatase Kit 3; Vector Laboratories,
Burlingame, CA). To maximize chromatic uniformity, sections were
stained with Masson trichrome and van Gieson stains in batches of 25
to 50 slides, and, for human SM ␣-actin, in batches of five slides, with
fresh reagents used for each specimen.
Analysis
Whole stained histologic sections were imaged in color with one of
two digital cameras equipped with a 50-mm fixed or variable macro
lens (Nikon, Tokyo, Japan). One camera (Leaf Lumina; ScyTech, Bedford, MA) had a resolution of 3400 ⫻ 2800 pixels, and the other (D1X;
Nikon) had a resolution of 3008 ⫻ 1960 pixels in 24-bit color. Images
were spatially calibrated by imaging a 1-mm rectilinear grid affixed to
a glass slide before and after each session of histologic imaging. Higherpower images were obtained by mounting either digital camera on a
microscope (BH-2; Olympus, Lake Success, NY).
Coronal MRIs of each specimen were used as the spatial standard to
correct histologic images for shrinkage and nonuniform distortions
introduced by later processing. For this purpose, one coronal plane
MRI was selected for each specimen from the level of the globe
equator, so that it included the rectus EOMs and pulleys (Fig. 1, top
left).17 A more anterior MRI plane would have better intersected the
densest pulley regions, but the flatness of EOM tendons and the density
of connective tissue make it difficult to distinguish contours required
to judge EOM or pulley position in the image plane.16 First, MRIs in
NIH Image were absolutely scaled, based on the calibrated MRI field of
view. Then, the MRI and histologic images (Fig. 1, center left) were
superimposed at partial transparency in image management software
(Photoshop, ver. 5.5; Adobe Systems, San Jose, CA) and rotated to
identical orientations (Fig. 1, bottom left). To correct for shrinkage, the
histologic images were then differentially scaled in the horizontal and
vertical directions so that the orbital walls superimposed on the MRI
(Fig. 1, bottom left). This procedure assumes that dimensional distortion produced by shrinkage or stretching affects the entire section
similarly in the same direction, but allows this distortion to be different
in orthogonal directions. For example, sections tended to stretch in the
dimension parallel to the motion of the microtome knife, but did not
do so perpendicular to this direction. Specimens were not uniformly
oriented in relation to the direction of knife travel, and shrinkage was
therefore represented for the two arbitrary orthogonal directions required for superimposition of histologic images on MRIs. Shrinkage in
orthogonal directions ranged from 30% ⫻ 40% (smaller ⫻ larger dimension) in the exenterated specimen to 10% ⫻ 15% in specimens
processed with orbital bones intact, with mean (⫾SD) shrinkage of
20% ⫾ 9% ⫻ 27% ⫾ 7%. All dimensional measurements reported
herein were performed with images individually corrected for bidirectional shrinkage distortions.
Two high-quality sections, corresponding to vicinity of the MRI
plane, were selected for measurement of pulleys and connective tissues defined as in Figure 1. It must be noted that the quasicoronal
plane perpendicular to the orbital axis, although optimal for all the
rectus EOMs in aggregate, is not exactly perpendicular to any rectus
EOM or parallel to the planes of the rectus pulleys. To measure the
thickness of the orbital and global aspects of the rectus pulleys, it was
necessary to identify the thickest regions using different sections.
Thicknesses of pulleys and connective tissue and SM bands were
measured from the digital images using the measurement tool. Orbital
and global thickness of each of the six EOM pulleys, the band connecting the medial rectus (MR) and superior rectus (SR) pulleys
(MR–SR band), the band connecting the MR and inferior rectus (IR)
pulleys (MR–IR band), and the band connecting the lateral rectus (LR)
and SR pulleys (LR–SR band) were determined in duplicate in two
sections for each specimen. Each thickness determination consisted of
the mean (⫾SE) of seven measurements distributed in representative
fashion over the extent of the structure.
Sections stained with Masson trichrome were used to measure
collagen density and pulley thickness. Densitometry was performed
IOVS, September 2002, Vol. 43, No. 9
Human Pulley Structure
2925
FIGURE 1. Correction of stretching
distortions produced by histologic
handling using preprocessing MRI.
Sample coronal section at level of
pulleys from 17-month-old human
right orbit. Left: coronal MRI of the
sample with bony orbit intact (top);
low-power coronal micrograph
stained with Masson trichrome (middle); dimensions and shape of the
micrograph were digitally corrected
to superimpose with the comparable
coronal MRI as shown at partial transparency (bottom). Enlarged and labeled view of corrected micrograph
at right. Rectangles A, B, and C are
magnified in Figure 3.
digitally using the color range command in the image management
software (Photoshop; Adobe). Collagen, constituting the most abundant and anatomically defining constituent of pulleys, stains blue with
Masson trichrome. Most pulleys and connective tissue bands consist of
laminae of collagen separated by voids that probably contain adipose
tissue. The relative proportion of pulley or band thickness that contained collagen, as opposed to void, was considered to be collagen
density. Collagen density was measured in the same sections stained
with Masson trichrome and at the same low magnification used for
measurement of pulley thickness. Density of the less-abundant microscopic elastin fibrils, which stain dark black with van Gieson elastin
FIGURE 2. Adjacent 10-␮m coronal
sections of 17-month-old human. (A)
Masson trichrome stain showing
dense collagen (dark blue) of MR
pulley ring. (B) Van Gieson elastin
stains appears black in region of
dense deposition of elastin. Elastin
fibrils, with larger aggregates denoted by arrows, are not individually
resolvable at this low magnification,
but are seen at higher power in Figure 3E. (C) Immunostaining for human SM ␣-actin (blue).
stain (Fig. 2B), was determined from high-power digital micrographs of
adjacent sections. SM density was determined from high-power digital
micrographs of adjacent sections immunostained with a blue chromogen for human SM ␣-actin, which is highly specific for SM (Fig. 2C). For
each connective tissue constituent, the selected color was converted
to grayscale. Grayscale images of specific tissue constituents were
imported into NIH Image, and the density function was applied with
the use of a consistent threshold. Density (percentage) was calculated
as pixels exceeding threshold divided by grayscale pixels. Use of a
threshold is unavoidable for quantitative analysis, because chromatic
variation among specimens is inevitable, even with large-scale histo-
2926
Kono et al.
IOVS, September 2002, Vol. 43, No. 9
FIGURE 3. Magnified views from
field in Figure 1 showing adjacent
10-␮m coronal sections of 17-monthold human orbit with columns illustrating the MR–SR band (Fig. 1A), the
MR pulley (Fig. 1B), and the MR–IR
band (Fig. 1C). (A–C) Masson
trichrome stain demonstrating dense
collagen (dark blue) in all regions.
(D–F) van Gieson elastin stain demonstrating elastin fibers (black)
richer in the orbital aspects of the
MR pulley and MR–IR band, than in
MR–SR band. (G–I) Immunoreactivity (blue) for human SM ␣-actin
showing small and large SM bundles
at the MR pulley and MR–IR band,
but only small bundles in the MR–SR
band.
logic processing techniques such as those used in the current study.
Use of a threshold that appears reasonable in comparison with the
original color images and is qualitatively consistent among specimens
minimizes the effects of threshold selection. Identifiable blood vessels
were excluded from areas of SM measurement, because vascular SM is
not relevant to mechanical measurements. We estimated total connective tissue content to be the average density multiplied by average
thickness of the structure under consideration.
A distinct band of connective tissue with an insertion on orbital
bone was considered to constitute an enthesis. Total collagen and
elastin content at the level of each enthesis was also measured using
the same methods. The dense connective tissue extending from the
MR pulley to the medial orbital periosteum was considered the medial
enthesis. The dense connective tissue band extending from the LR
pulley to the lateral periosteum was considered to be the lateral
enthesis, distinguishable from the enthesis of the lateral canthal tendon, slightly more anterior on the periosteum of the zygomatic bone.
The anteroposterior extent of the MR–IR SM band was measured in
each specimen, as judged by the maintenance of the band structure in
serial 10-␮m-thick sections. Data on thickness and extent in each
specimen were normalized to that specimen’s globe diameter compared with a normal globe diameter in vivo of 24.3 mm obtained by
MRI.37 Enthesis dimensions were also measured. The SO pulleys
(trochleas) in two specimens (57 and 93 years old) were not measured
because of damage sustained during specimen preparation.
RESULTS
Major histologic findings were similar in all specimens, ranging
in age from fetal to 93 years. In low-power micrographs near
the globe equator where the pulley ring was present, Masson
trichrome stain clearly distinguished the EOMs (purple) surrounded by pulleys consisting of encirclements of collagen
(Fig. 1, blue). Structural details are clearer at higher power (Fig.
2), showing adjacent 10-␮m coronal sections of a 17-month-old
human MR stained with Masson trichrome stain (Figs. 2A, 3B)
and van Gieson elastin stain (Figs. 2B, 3E) and for human SM
␣-actin (Figs. 2C, 3H). The image plane in Figure 2 intersects a
dense region of the orbital aspect of the MR pulley, showing a
dense encirclement by collagen laminae having only sparse
voids that presumably contained fat before elution in process-
ing. More posterior sections demonstrated a thicker ring on the
global aspect of the MR pulley, but the global aspect was not as
thick as the orbital. Black elastin fibrils were embedded in the
collagen of the pulley ring and in suspensory bands running
superiorly and inferiorly (Figs. 2B, 3). SM, appearing blue due
to immunoreactivity for human SM ␣-actin, was abundant on
the orbital surface of the MR pulley (Figs. 2C, 3). The orbital
layers of the rectus EOMs inserted into the collagen of each of
their respective pulleys in every specimen. The morphology of
the MR pulley was identical in all specimens except the fetal
one.
Suspensory bands of connective tissue were also consistently identified in all specimens (Fig. 1). A dense band was
present from the MR to the IR pulleys (MR–IR band, Fig. 1), and
from the MR to the SR pulleys (MR–SR band, Fig. 1). Another
dense band was present from the SR to the LR pulleys (LR–SR
band, Fig. 1). The IO muscle was present between the IR and
LR pulleys. The collagenous sheath of the IO was contiguous
with the LR and IR pulleys, forming an interconnection analogous to the other interconnections of contiguous pulleys. Anterior to the pulley rings, posterior Tenon fascia formed a
complete cup investing the globe analogous to a ball within a
socket, and inserting most anteriorly on the orbital rim. The
posterior Tenon fascia contained abundant collagen and elastin.
Quantitative analysis recognized that, because of the intersection of coronal histologic sections with the three-dimensional structure of the orbital connective tissues, the coronal
histologic planes would not exactly coincide with the plane of
each pulley ring. Total pulley thickness for each EOM was
measured separately for the orbital and global portions, in each
case from the coronal section showing the most complete
development of that portion. The tissue bands interconnecting
the pulleys were measured from sections showing maximal
development of the two pulleys under consideration. In each
case, measurements included all connective tissue constituents, plus intervening spaces that had presumably contained
fat (Fig. 4). The thickness, collagen, and elastin contents in the
three specimens aged 17 months, 4 years, and 57 years were
similar and were averaged to represent data from the predominant part of the postnatal human age range, whereas quanti-
IOVS, September 2002, Vol. 43, No. 9
FIGURE 4. Connective tissue thickness normalized to standard global
diameter. Data are the average of the three specimens aged 17 months,
4 years, and 57 years. *Not measured in the 57- and 93-year-old specimens, because of exenteration damage to the trochlea.
tative data from the fetal and 93-year-old specimens were
considered separately. Pulley thickness tended to be greater on
the orbital than the global aspect (Fig. 4). For the mid–agerange specimens, mean pulley thickness of the orbital aspect
ranged from 1.0 to 2.25 mm, whereas global aspect thickness
ranged from 0.5 to 1.75 mm. Pulley thickness was modestly
lower in the fetal and 93-year-old specimens. Interconnections
among pulleys ranged in mean thickness from 1.25 mm for the
MR–SR band to 2.75 mm for the MR–IR band. Similar to the
thickness of the pulleys, interconnecting band thickness was
generally lower in the fetal and 93-year-old specimens than in
the other specimens. The SO sheaths were thinner than in the
remaining EOMs.
Total collagen content was taken to be the product of
pulley thickness and collagen density, and its dimension was
thus expressed in millimeters (Fig. 5). The orbital part of each
rectus pulley had a mean collagen content of 0.5 to 0.75 mm in
the three specimens aged 17 months, 4 years, and 57 years.
The fetal specimen had collagen content similar to these, but
the 93-year-old specimen consistently had less collagen. The
global part of each rectus pulley had mean collagen content of
0.25 to 0.5 mm, again with the 93-year-old specimen having
less. The MR–IR band had the greatest collagen content of any
structure studied, averaging 1 mm.
Total elastin content was taken to be the product of pulley
thickness and elastin density and its dimension was also expressed in millimeters (Fig. 6). Elastin content was much less
than collagen, ranging downward from a mean of approximately 0.075 mm in the MR pulley orbital layer and MR–IR
band in the three specimens aged 17 months, 4 years, and 57
years. Unlike collagen, elastin content was often greater in the
93-year-old specimen. Elastin fibrils were qualitatively different
in the oldest specimen, showing evidence of shredding and
clumping. Elastin content was almost uniformly zero in the
fetal specimen, except in the SO sheath where elastin was
present.
Because of careful preparation of whole orbits processed en
bloc in continuity with orbital bones, it was possible to study
the structure of each major connective tissue enthesis on the
periosteum (Fig. 7). This included all except the 93-year-old
specimen, which was processed by exenteration before fixation and thus had dissection damage to the entheses. The
Human Pulley Structure
2927
FIGURE 5. Total collagen content computed as mean pulley thickness
multiplied by mean collagen density, normalized by standard global
diameter. Data are the average of the three specimens aged 17 months,
4 years, and 57 years. *Not measured in the 57- and 93-year-old specimens, because of exenteration damage to the trochlea.
pulley suspensory system entheses corresponded to the horizontal rectus pulleys: laterally on the zygomatic bone at what is
probably the zygomatic tubercle,38 medially at the lacrimal,
and probably also the maxillary and frontal bones. These entheses have been described as the “check ligaments,” and it has
been correctly recognized that only the MR and LR are endowed with them.38 The lateral enthesis extended directly to
the LR pulley, which was then indirectly coupled to the SR
through the LR–SR band, and indirectly coupled to the IR along
the IO muscle and its sheath. The lateral enthesis was distinct
from the lateral canthal tendon extending through the lacrimal
grand and having its own enthesis on the zygoma at the orbital
rim approximately 3 mm more anteriorly. The lateral enthesis
FIGURE 6. Total elastin content computed as mean pulley thickness
multiplied by mean elastin density, normalized by standard global
diameter. Data are the average of the three specimens aged 17 months,
4 years and 57 years. *Not measured in the 57- and 93-year-old specimens because of exenteration damage to the trochlea. There was
greater elastin in the orbital than the global aspect of the MR pulley and
rich elastin in the MR–IR band.
2928
Kono et al.
IOVS, September 2002, Vol. 43, No. 9
FIGURE 7. Coronal section of 17month-old old human right orbit at
the level of the entheses, stained
with Masson trichrome. Residual
bone stained bright red, but was
largely removed from the underlying
periorbita after fixation, to enable
sectioning. Dense, blue-staining collagen is visible in the connective tissues. Left: high-power view at the
level of lateral canthal tendon approximately 4 to 5 mm anterior to
lateral enthesis (top) and high-power
view of lateral enthesis extending
from the orbital rim at zygomatic
bone to the LR pulley (bottom).
Sheath of SO tendon unites with nasal aspect of the SR pulley.
measured approximately 5 mm mediolaterally and anteroposteriorly and 6 mm dorsoventrally (Fig. 7) and often incorporated a central lobule of lacrimal gland that was unavoidably
included in these dimensions. Dimensions of the lateral enthesis for each specimen are presented in Table 1. The medial
enthesis extended directly to the MR pulley, which was then
coupled indirectly to the SR and IR through their corresponding bands. The medial enthesis measured approximately 10
mm in mediolateral and 11 mm in dorsoventral extent, but only
4 mm in anteroposterior extent. Measured dimensions for the
medial enthesis of each specimen are presented in Table 1,
except for the 57-year-old specimen, in which processing damage prevented accurate determination of the anteroposterior
extent. The medial enthesis band incorporated a slip of anteroposteriorly oriented striated muscle in its center (Fig. 7) that
was unavoidably incorporated in its measured dimensions (Table 1). The medial canthal tendon had its enthesis at the same
site as that of the band to the MR. There were no direct
entheses from the vertical rectus pulleys to the immediately
adjacent orbital bones.
The pulley enthesis bands were composed of very dense
connective tissue. Total collagen content of both the medial
and lateral enthesis bands exceeded that of any other pulley
suspensory structure, whereas total elastin content was comparable to that of other dense pulley suspensions. Dimensions
of enthesis bands in individual specimens are reported in
Table 2.
Nonvascular SM was stereotypically but not uniformly distributed in the orbital connective tissues of all specimens, and
was identified by immunoreactivity to human SM ␣-actin. Recognizable blood vessels were avoided in the quantitative evaluations, but a few small vessels may have been included if their
lumina could not be identified. There were two types of SM
distributions (Fig. 3). Large SM bundles averaged 30 to 40 ␮m
in diameter and were mainly located in a band from the
superior border of the MR to the nasal border of the IR. These
large bundles were apparently cut transversely by the coronal
sections. Small SM bundles averaging 10 ␮m in diameter were
located on the global surface of the large cell bundles, but
formed a band having greater length, extending from the nasal
border of the SR to the nasal border of the IR. The small SM
bundles appeared to have been cut more tangentially, so that
their long axes approximated the coronal plane. Because it
formed a distinct structure (Figs. 3B, 3E, 3H), the SM band at
the MR pulley was not included in the pulley’s thickness, but
the SM bundles at the MR–IR band were included in the
pulley’s thickness, because the bundles were intrinsic within
the band (Figs. 3C, 3F, 3I). For quantitative analysis, large and
small SM bundle distributions were lumped together as a single
structure. The normalized anteroposterior extent of the MR–IR
TABLE 1. Pulley System Enthesis Band Dimensions
Mediolateral
Dorsoventral
Anteroposterior
Specimen Age
Medial
Enthesis
Lateral
Enthesis
Medial
Enthesis
Lateral
Enthesis
Medial
Enthesis
Lateral
Enthesis
33-wk fetus
17 mo
4y
57 y
Mean ⫾ SE
11.03
10.42
8.18
11.29
10.23 ⫾ 0.71
3.28
3.26
5.26
6.11
4.48 ⫾ 0.72
13.12
10.20
9.47
12.02
11.20 ⫾ 0.83
8.97
4.12
3.87
7.06
6.01 ⫾ 1.23
2.36
2.35
3.16
ND
2.62 ⫾ 0.27
5.11
3.49
3.97
6.93
4.88 ⫾ 0.77
Data are expressed in millimeters. ND, not determined.
Human Pulley Structure
IOVS, September 2002, Vol. 43, No. 9
2929
TABLE 2. Total Collagen and Elastin Content of Pulley System Enthesis Bands
Total Collagen Content (mm)
Total Elastin Content (mm)
Specimen Age
Medial
Enthesis
Lateral
Enthesis
Medial
Enthesis
Lateral
Enthesis
33-wk fetus
17 mo
4 mo
57 mo
Mean ⫾ SE
2.50
2.59
1.55
1.35
2.00 ⫾ 0.32
1.42
1.10
0.93
1.71
1.29 ⫾ 0.17
0.080
0.158
0.087
0.057
0.095 ⫾ 0.022
0.003
0.066
0.010
0.070
0.037 ⫾ 0.018
SM band of the five specimens averaged 4.74 ⫾ 0.82 mm (⫾SE;
range, 2.77– 6.68 mm). The LR–SR band, as well as other
connective tissue bands, had few and widely scattered small
SM cells. The MR–SR band and the LR–SR band had insufficient
SM for quantitative measurement by ␣-actin immunoreactivity.
The lateral levator aponeurosis (LLA) has been described to
be a connective tissue condensation extending between the
superior border of the LR pulley and the lateral border of the
levator palpebral superioris (LPS).19 Careful examination of the
our high-quality specimens suggested the need for revision of
this description. More correctly, the LLA should be considered
to be a lateral expansion of the LPS tendon, running inferolaterally and partially through the lacrimal gland to insert on the
orbital bone at the lateral canthus. The muscular tissues of the
LLA began superiorly as a lateral extension of striated muscle
from the LPS in the medial half, contiguous with a particularly
dense band of SM in the lateral half of the LLA. The SM in the
LLA did not appear to be directly coupled to the oculorotary
EOMs. However, another SM distribution, distinct from the
LR–SR band, extended from a lateral expanse of striated muscle
fibers from the levator and continued laterally and inferiorly to
the LR pulley. This distribution, forming part of what was
described by Müller as the “peribulbar muscle,”39,40 was
present anterior to the equator.
With both large and small bundles included, maximum SM
thickness near the MR pulley or MR–IR band ranged from 0.2
FIGURE 8. SM thickness normalized by standard adult globe diameter.
Only the MR pulley region and MR–IR band had sufficient SM for
quantitative analysis. SM in these regions was measured in sections
corresponding to those used for analysis of collagen and elastin. Maximum SM thickness was determined from examination of all sections
from each specimen. There was insufficient SM for measurement in the
MR–IR band in the 93-year-old specimen. Except for maxima, duplicate
measurements were made in the same orbits (duplicated symbols).
to 1.4 mm (Fig. 8). Total SM content was taken to be the
product of band thickness and SM density, thus having the
dimension of mm (Fig. 9). The SM content at the MR pulley in
the region of the medial enthesis ranged from 0.01 to 0.13 mm.
DISCUSSION
Recent formulations of EOM biomechanics emphasize the pivotal role of EOM pulleys in ocular kinematics.22,24,32,41,42
These formulations are supported by MRIs in living humans
that demonstrate the highly stereotyped locations and stability
of rectus pulleys.20,27,37 Qualitative features of EOM pulleys are
similar in humans, monkeys,19 rats,43 and rabbits (Demer JL,
unpublished data, 2002). Further support for the importance
of pulleys is provided here by the demonstration of quantitative histologic uniformity in composition and structural features of pulleys in specimens prepared en bloc to maintain
critical anatomic relationships. The general arrangement of
orbital connective tissues is uniform throughout the range of
human age from fetal life to the tenth decade and is summarized in schematic form in Figure 10. After fetal life when it was
smaller, even the normalized thickness of pulley rings and
suspensory bands remained high and roughly consistent across
the age range. Such uniformity supports the concept that
pulleys and orbital connective tissues constitute important
components of the ocular motor apparatus. Although precise
timing of pulley maturation cannot be determined from the
available specimens, the pulley system of even the 17-monthold specimen had adult characteristics. Only fetal pulley structure showed immaturity in development of collagen thickness
and elastin content (Figs. 4, 5, 6).
FIGURE 9. SM content computed as the mean thickness multiplied by
mean density, both determined from SM ␣-actin immunoreactivity, and
normalized by standard adult globe diameter. Sites and duplicate measurements denoted as in Figure 8.
2930
Kono et al.
FIGURE 10. Frontal view of major orbital connective tissues. Globe is
depicted as sectioned at approximately the coronal plane level of the
pulley rings, but the illustration also includes the more anteriorly
located medial and lateral entheses. LG, lacrimal gland; SOT, superior
oblique tendon.
Intercouplings between adjacent pulleys were also stereotypic in configuration and composition (Fig. 10). The MR–IR
band was not only the thickest such intercoupling, but it also
contained the most collagen, elastin, and SM. These features of
the MR–IR band seem ideal to provide a stiff elastic coupling
between the highly stable MR and relatively mobile IR pulleys,20 permitting the latter to move nasally and laterally, not
only under passive tension but also nasally during contraction
of the intrinsic band of small SM cells. This may be one
mechanism underlying nasal repositioning of the IR pulley
during convergence.31,32 In rat, a species without convergence, the MR–IR band is poorly developed and contains minimal SM.43 The MR–IR band forms part of the Lockwood
ligament, a structure that has been described to form a kind of
suspensory sling running mediolaterally inferior to the globe.38
However, the present study indicates that the connective tissue of the Lockwood ligament temporal to the IR pulley is
much less substantial than that nasal to it. The MR–SR band,
although thinner and containing less connective tissue and
much less SM, seems also sufficiently robust to provide significant mechanical coupling between these two pulleys. The
MR–SR band should be considered to have lower stiffness than
the MR–IR band.
Earlier studies of orbital anatomy were hampered by difficulties in removing the delicate, elastic, soft tissues from the
supporting orbital bones. Removal of the soft tissues by dissection almost inevitably results in damage, particularly to the
entheses, where adhesion to bone is most secure. In the current study, we used specimens that, for the most part, had
been processed in continuity with the orbital bones to clarify
the relationship between the pulley system and its bony support, as depicted in the schematic frontal view in Figure 10.
This schematic depicts only the thick and presumably mechanically significant connective tissue structures, omitting the thinner anterior and posterior pulley slings. Figure 10 also depicts
structures that generally do not all lie in the same plane.
Contrary to the depiction in an earlier schematic, the IR and SR
pulleys were found not to have direct mechanical coupling to
an enthesis on the adjacent orbital bone. The rectus pulley
system is mechanically coupled to the anterior orbital bones by
heavy connective tissues nasally and temporally. The medial
enthesis corresponds to the medial canthal tendon region
where the lids are anchored nasally, whereas the lateral enthesis is several millimeters posterior to the lateral canthal tendon
IOVS, September 2002, Vol. 43, No. 9
at the zygomatic tubercle. The absence of direct enthesis of the
vertical rectus pulleys to the bone of the adjacent orbital rim
regions is a necessary consequence of the mobility of the
superior and inferior eyelids, which move through these regions within only loose connective tissues. The vertical rectus
pulleys are thus indirectly coupled to the medial and lateral
entheses by the intercouplings between the rectus pulleys.
Further stability is provided by the SO tendon emerging from
its rigid trochlea and by the sheath and belly of the IO muscle,
which originates from anterior orbital bone.
Past descriptions suggested correctly that the orbital connective tissue system supports and protects the globe, but such
descriptions emphasized a major role for the “check ligaments”
in limiting and dampening ocular movements.38 The elasticity
of the check ligaments was believed to be important in graduating the action of EOM contraction to ensure smooth ocular
rotations without jolting the globe.38 The active-pulley hypothesis,22,32 as well as the current quantitative anatomic findings,
support quite a different interpretation of the check ligaments— elastic suspensions of the pulley system that actively
regulates the direction of EOM force to control ocular kinematics. In view of the misleading functional connotation, the term
check ligament should probably be abandoned.
The current report also makes a novel distinction between
the LLA and the LR–SR band. The LLA, connecting the lateral
expansion of the LPS to the anterior bony orbit while partially
traversing the lacrimal gland, appears to be mainly a suspension for the LPS in relationship to the Whitnall ligament. The
LLA would only indirectly stabilize the SR pulley by mechanical
coupling of the LPS and SR. The lateral half of the LLA contains
abundant SM, whereas the medial half contains striated muscle
contiguous with the LPS. In contrast, the LR–SR band extends
directly between the involved pulleys and contains only a small
amount of SM in small bundles. The LR–SR band is equivalent
to the MR–SR band in thickness and collagen content, but
probably has lower stiffness, because the elastin content of the
LR–SR band is about one fourth of the MR–SR band. Elastin has
the property of reversible extensibility44 that probably confers
elastic stiffness on pulley suspensory tissues.
The orbital aspect of each rectus pulley had more and
thicker collagen than the global aspect. Elastin content in the
orbital aspect was also richer than in the global aspect of each
pulley. The more extensive and presumably stiffer structure of
orbital aspects of rectus pulleys is likely to be related to concentration at those sites of stress associated with sharp EOM
path inflections in secondary and tertiary gaze positions.27,37
Such inflections, which move in the anteroposterior direction
during contraction of orbital layer fibers inserting on the pulleys,22 require support from pulley suspensions to the entheses
to maintain their considerable resistance to sideslip in the
coronal direction.
Although general features of pulleys were preserved in the
93-year-old specimen, there was also qualitative evidence of
age-related degeneration. Elastin fibers in pulley showed
clumping and shredding. The change may account for the
observation that total elastin thickness in the 93-year-old specimen exceeded that of the other specimens (Fig. 6). There also
was obvious qualitative atrophy of collagen fibers in the 93year-old specimen, associated with reduced total collagen
thickness (Fig. 5). Degenerative connective tissue changes in
aging correlate with limited ocular ductions in the elderly,45
and with MRI evidence of downward displacement of horizontal rectus pulley positions.46 Asymmetrical occurrence of such
changes in the two orbits may be expected to cause strabismus.46
SM has long been recognized in the orbital connective
tissues,5,19,39 but its role has been unclear. Demer et al.19
described an intricate innervation pattern, including rich sym-
IOVS, September 2002, Vol. 43, No. 9
pathetic, parasympathetic, and nitroxidergic innervation to
pulley SM, suggesting the following possible roles for pulley
SM: to maintain uniform stiffness in the pulley suspensions; to
accomplish slow, adaptive adjustments in pulley locations as
are necessary to maintain binocular alignment over a lifetime;
and to fulfill a possible dynamic role in eye movements. The
latter suggestion is consistent with new observations of excyclorotation of the four-rectus pulley array during convergence.31 The especially dense peribulbar SM between the IR
and MR pulleys seems anatomically suited to accomplish the
observed nasal shifting of the IR pulley in convergence.
Mathematical models of EOMs have led to much insight into
EOM function, and have motivated significant new lines of
inquiry.47 However, even the most comprehensive currently
available models34 fail to account for important features such
as large gaze-related shifts in rectus pulley positions.27 Quantitative data are now available for each of the four rectus EOMs,
showing the number of global layer fibers inserting on the
scleral tendon and the number of orbital layer fibers inserting
on the pulley.21 More complete models may incorporate this
and the current quantitative data on pulley structure and interconnections in computational implementations of these critical structures. The values for collagen and elastin content
reported here provide a basis for reasonable estimates of relative stiffness of pulley suspensions and interconnections. Potential for pulley shift due to the actions of striated EOMs may
be estimated from connective tissue and SM content in the
specific structures described herein. A realistic computational
model of EOM and orbital connective tissue mechanics would
be of general value in understanding normal and pathologic
behavior of ocular movements and binocular alignment.
Acknowledgments
The authors thank Nicolasa de Salles and Frank Henriquez for technical
assistance.
References
1. Lockwood CB. The anatomy of the muscles, ligaments, and fasciae,
etc. J Anat Physiol. 1886;20:1–25.
2. Bergen MP. Some histological aspects of the structure of the
connective tissue system and its relationships with the blood
vessels in the human orbit. Acta Morphol Neerl Scand. 1982;20:
293–308.
3. Jones LT. A new concept of the orbital fascia and rectus muscle
sheaths and its surgical implications. Trans Am Acad Ophthalmol
Otolaryngol. 1968;72:755–764.
4. Koornneef L. Details of the orbital connective tissue system in the
adult. Acta Morphol Neerl Scand. 1977;15:1–34.
5. Sappey PC. Traite D’Anatomie Desccriptive Avec Figures Intercalees Dans Le Texte. Paris: Delahaye et Lecrosnier; 1888.
6. Koornneef L. Spatial Aspects of Orbital Musculo-fibrous Tissue in
Man. Amsterdam: Swets & Zeitlinger; 1977.
7. Koornneef L. The architecture of the musculo-fibrous apparatus in
the human orbit. Acta Morphol Neerl Scand. 1977;15:35– 64.
8. Koornneef L. New insights in the human orbital connective tissue:
result of a new anatomical approach. Arch Ophthalmol. 1977;95:
1269 –1273.
9. Koornneef L. Orbital septa: anatomy and function. Ophthalmology. 1979;86:876 – 880.
10. Robinson DA. A quantitative analysis of extraocular muscle cooperation and squint. Invest Ophthalmol. 1975;14:801– 825.
11. Miller JM, Robinson DA. A model of the mechanics of binocular
alignment. Comput Biomed Res. 1984;17:436 – 470.
12. Miller JM, Robins D. Extraocular muscle sideslip and orbital geometry in monkeys. Vision Res. 1987;27:381–392.
13. Simonsz HJ, Harting F, de Waal BJ, Verbeeten BWJM. Sideways
displacement and curved path of recti eye muscles. Arch Ophthalmol. 1985;103:124 –128.
Human Pulley Structure
2931
14. Miller JM. Functional anatomy of normal human rectus muscles.
Vision Res. 1989;29:223–240.
15. Miller JM, Demer JL, Rosenbaum AL. Effect of transposition surgery
on rectus muscle paths by magnetic resonance imaging. Ophthalmology. 1993;100:475– 487.
16. Clark RA, Rosenbaum AL, Demer JL. Magnetic resonance imaging
after surgical transposition defines the anteroposterior location of
the rectus muscle pulleys. J Am Acad Pediatr Ophthalmol Strabismus. 1999;3:9 –14.
17. Demer JL, Miller JM, Poukens V, Vinters HV, Glasgow BJ. Evidence
for fibromuscular pulleys of the recti extraocular muscles. Invest
Ophthalmol Vis Sci. 1995;36:1125–1136.
18. Porter JD, Poukens V, Baker RS, Demer JL. Structure-function
correlations in the human medial rectus extraocular muscle pulleys. Invest Ophthalmol Vis Sci. 1996;37:468 – 472.
19. Demer JL, Poukens V, Miller JM, Micevych P. Innervation of extraocular pulley smooth muscle in monkeys and humans. Invest
Ophthalmol Vis Sci. 1997;38:1774 –1785.
20. Clark RA, Miller JM, Demer JL. Location and stability of rectus
muscle pulleys inferred from muscle paths. Invest Ophthalmol Vis
Sci. 1997;38:227–240.
21. Oh SY, Poukens V, Demer JL. Quantitative analysis of extraocular
muscle global and orbital layers in monkey and human. Invest
Ophthalmol Vis Sci. 2001;42:10 –16.
22. Demer JL, Oh SY, Poukens V. Evidence for active control of rectus
extraocular muscle pulleys. Invest Ophthalmol Vis Sci. 2000;41:
1280 –1290.
23. Haslwanter T. Mathematics of three-dimensional eye rotations.
Vision Res. 1995;35:1727–1739.
24. Quaia C, Optican LM. Commutative saccadic generator is sufficient
to control a 3-D ocular plant with pulleys. J Neurophysiol. 1998;
79:3197–3215.
25. Tweed D, Vilis T. Geometric relations of eye position and velocity
vectors during saccades. Vision Res. 1990;30:111–127.
26. Ruete CGT. Ocular physiology. Strabismus. 1999;7:43– 60.
27. Kono R, Clark RA, Demer JL. Active pulleys: magnetic resonance
imaging of rectus muscle paths in tertiary gazes. Invest Ophthalmol Vis Sci. 2002;43:2179 –2188.
28. Demer JL, Clark RA, Miller JL. Magnetic resonance imaging (MRI)
of the functional anatomy of the inferior oblique (IO) muscle
[ARVO Abstract]. Invest Ophthalmol Vis Sci. 1999;40(4):S772.
Abstract nr 4073.
29. Demer JL, Oh SY, Poukens V. The orbital layers of human and
monkey oblique extraocular muscles (EOMs) insert on the orbital
connective tissue system [ARVO Abstract]. Invest Ophthalmol Vis
Sci 2001;42(4):S517. Abstract nr 2781.
30. Demer JL. Mechanical interactions of oblique extraocular muscles
(EOMs) with actively controlled rectus pulleys maintain kinematics of linear oculomotor plant. Soc Neurosci Abstr. 2001;27:2071.
31. Demer JL, Kono R, Wright W. Magnetic resonance imaging (MRI)
of human extraocular muscles (EOMs) during asymmetrical convergence [abstract]. 2002 Annual Meeting Abstract and Program
Planner accessed at www.arvo.org. Association for Research in
Vision and Ophthalmology. Abstract 1915.
32. Demer JL. The orbital pulley system: a revolution in concepts of
orbital anatomy. Ann NY Acad Sci. 2002;956:17–32.
33. Miller JM, Bockisch CJ, Pavlovski DS. Missing lateral rectus force
and absence of medial rectus co-contraction in ocular convergence. J Neurophysiol. 2002;87:2421–2433.
34. Miller JM, Pavlovski DS, Shaemeva I. Orbit 1.8 Gaze Mechanics
Simulation. San Francisco: Eidactics; 1999.
35. Porrill J, Warren PA, Dean P. A simple control laws generates
Listing’s positions in a detailed model of the extraocular muscle
system. Vision Res. 2000;40:3743–3758.
36. Sheehan DC, Hrapchak BB. Theory and Practice of Histotechnology. St. Louis: Mosby; 1973:95–116.
37. Clark RA, Miller JM, Demer JL. Three-dimensional location of
human rectus pulleys by path inflections in secondary gaze positions. Invest Ophthalmol Vis Sci. 2000;41:3787–3797.
38. von Noorden GK. Binocular Vision and Ocular Motility: Theory
and Management of Strabismus. St. Louis: Mosby; 1996:45–50.
2932
Kono et al.
39. Müller H. Uber einen glatten Muskel in der Augenhohle des Menschen and dur Saugethiere. Zwiss Zool. 1858;9:541.
40. Page RE. The distribution and innervation of the extraocular
smooth muscle in the orbit of the rat. Acta Anat. 1973;85:10 –18.
41. Demer JL. Extraocular muscles. In: Jaeger EA, Tasman PR, ed.
Clinical Ophthalmology. Vol. 1. Philadelphia: Lippincott; 2000:
chap 1.
42. Raphan T. Modeling control of eye orientation in three dimensions. I: role of muscle pulleys in determining saccadic trajectory.
J Neurophysiol. 1998;79:2653–2667.
43. Khanna S, Porter JD. Evidence for rectus extraocular muscle pulleys in rodents. Invest Ophthalmol Vis Sci. 2001;42:1986 –1992.
IOVS, September 2002, Vol. 43, No. 9
44. Christiano AM, Uitto J. Molecular pathology of the elastic fibers.
J Invest Dermatol. 1994;103:53S–57S.
45. Clark RA, Isenberg SJ. The range of ocular movements decreases
with aging. J Am Acad Pediatr Ophthalmol Strabismus. 2001;5:
26 –30.
46. Clark RA, Demer JL. Effect of aging on human rectus extraocular
muscle paths demonstrated by magnetic resonance imaging. Am J
Ophthalmol. In press.
47. Miller JM, Demer JL. Biomechanical modeling in strabismus surgery. In: Rosenbaum AL, Santiago P, eds. Clinical Strabismus
Management: Principles and Techniques. St. Louis: Mosby; 1999:
99 –113.