Evidence for a Pulley of the Inferior Oblique Muscle

Abstract

purpose. This study was undertaken to investigate evidence for a connective tissue pulley constraining the path of the inferior oblique (IO) muscle.

methods. From magnetic resonance images, the cross-sectional area, path, and orbital relationships of the human IO were determined in multiple gaze positions. Rectus pulleys were directly imaged with intravenous gadodiamide contrast. Images were compared with serial histologic sections of IO muscles of humans and monkeys.

results. The IO path from origin to the lateral border of the inferior rectus (IR) muscle was straight. Lateral to the IR, the IO curved to follow the globe. At the point of IR crossing, the IO moved anteriorly from infraduction to supraduction by approximately 53% of the IR insertion’s travel. Gaze-related change in IO cross section was demonstrable near the IR center. The gaze-related inflection in IO path corresponded to its encirclement by a pulley consisting of a dense ring of collagen, stiffened by elastin and smooth muscle, and united with the IR pulley. Orbital layer fibers of the IO inserted on its pulley, the lateral rectus (LR) pulley, and associated connective tissues.

conclusions. Like the rectus muscles, the human and monkey IO has a connective tissue pulley serving as its functional origin. The position of the IO pulley is influenced by its coupling to the actively moving IR pulley, whereas in turn the IO orbital layer inserts on and presumably shifts the IR and LR pulleys. These intercouplings facilitate implementation by rectus extraocular muscle suspensions of a commutative ocular motor plant.

The rectus extraocular muscles (EOMs) do not pull toward their classic anatomic origins in the annulus of Zinn. Rather, EOM force is directed toward a much more anteriorly located, and mechanically stable, connective tissue structure called a pulley. ¹ Each rectus pulley consists of an encircling sleeve and ring located near the globe equator in Tenon’s fascia, ¹ coupled to the orbital wall, adjacent EOMs, and equatorial Tenon’s fascia by bands containing connective tissue and smooth muscle (SM). ² Pulleys inflect rectus EOM paths similarly to the inflection of the superior oblique (SO) tendon path by the trochlea. Magnetic resonance imaging (MRI) shows that the locations of the rectus pulleys in normal eyes are highly uniform. ³ Rectus EOMs have direct mechanical actions on their pulleys. The global layer (GL) of each rectus EOM, containing about half of the total EOM fibers, ⁴ passes through the pulley and becomes contiguous with tendon to insert on the globe. The orbital layer (OL), containing the remaining roughly half of the EOM fibers, inserts on the pulley. ^{4 5}  

Rectus pulleys are fundamental to ocular rotational properties, known as kinematics. With the head upright and immobile, torsional eye movement is described by a relationship known as Listing’s law. Listing’s law states that any eye orientation can be reached by rotation about an axis lying in a single plane, Listing’s plane. ⁶ An equivalent statement is that the eye’s rotational axis shifts by half of the shift in ocular orientation ⁷ and is called Listing’s “half-angle rule.” The important advantage of half-angle behavior is that it implements a linear ocular motor plant that is effectively commutative with respect to ocular rotations. ⁸ An anatomic basis for the half-angle rule based on rectus pulley movement was recently formulated in the active pulley hypothesis (APH). ^{5 9 10 11} The coordinated control postulate of the APH accounts for half-angle kinematics by supposing pulley movements in the orbit along the EOM axis that maintain a constant geometric relationship between the pulley relative and the scleral insertion. This behavior of rectus pulleys has been confirmed quantitatively by MRI in humans. ⁹  

The rectus EOMs alone seem capable of implementing Listing’s half-angle kinematics, ¹² so what might be the necessity for the oblique EOMs? The differential control postulate of the APH was proposed to account for ocular motor kinematics, such as observed during the vestibuloocular reflex (VOR) ¹³ and during convergence, ^{14 15 16} that do not conform to the classic description of Listing’s law. The differential control postulate proposed that differential innervation in rectus OLs and GLs, oblique EOM OLs, or orbital SM, separately may move the rectus pulleys differently from the travel of their scleral insertions ⁵ or even in directions transverse to the EOM axis (Demer JL, et al. IOVS 2002;43:ARVO E-Abstract 1915; Refs. ^{11 17} ) to alter ocular kinematics. The original supposition of differential control of rectus pulleys was of larger anteroposterior shifts during the VOR than during visually guided eye movements. ⁵ Although this original proposal no longer appears tenable for the horizontal and vertical VOR, ¹⁸ pulley repositioning in another direction, transverse to the rectus EOM paths, occurs during convergence (Demer JL, et al. IOVS 2002;43:ARVO E-Abstract 1915; Ref. ¹¹ ) and during the static torsional VOR. 

Histologic evidence exists for mechanical intercouplings among pulleys ¹⁹ and is supported by MRI showing stereotypic shifts of rectus pulleys during gaze shifts. ¹⁷ In particular, the IR pulley shifts nasally in supraduction, and temporally in infraduction, suggestive of possible mechanical coupling to the IO. ³ The dense neurofibrovascular bundle (NFVB) is located near the IR pulley and has been proposed to act as an ancillary origin for the IO after surgical anteriorization of its insertion. ²⁰ By MRI the IO muscle’s anterior border can be seen to shift posteriorly with infraduction by an amount similar to the motion of the inferior lid margin but less than the motion of the corneal limbus. ²¹ These findings suggest connective tissue linkages between the IO, IR, and lower eyelid. Like the rectus EOMs, ⁵ the IO consists of both a GL and OL. ²² Although there is now good evidence that the rectus OLs insert on and position the rectus pulleys, the insertion and actions of the IO’s GL remains enigmatic. The present study was conducted to determine whether, like the rectus EOMs, the IO has a soft pulley whose location is influenced by its OL. An additional purpose was to define the anatomic relationships of the OL of the IO to adjacent structures—in particular, the IR and lateral rectus (LR) pulleys. 

Methods

Magnetic Resonance Imaging

MRI was performed to define orbital anatomy in living humans. High-resolution, T1-weighted images were collected from volunteers who gave written informed consent to a protocol conforming to the Declaration of Helsinki and approved by the Institutional Review Board at the University of California, Los Angeles. All subjects underwent complete ophthalmic examination to verify normal corrected visual acuity, ocular motility, binocular alignment, anterior segment anatomy, and ophthalmoscopy. Imaging was performed with a 1.5-T scanner (Signa; General Electric, Milwaukee, WI). Crucial aspects of this technique, described in detail elsewhere, include use of a facemask-mounted, dual-phased surface coil array (Medical Advances, Milwaukee, WI), to improve signal-to-noise ratio, and fixation targets to avoid eye motion artifacts. ^{17 23} The head was stabilized with the volunteer supine. An array of illuminated fixation targets was secured in front of each orbit with the center target in subjective central position for each eye and in selected cases in secondary and tertiary gaze positions. Images of 2-mm thickness in a matrix of 256 × 256 were obtained over a field of view of 8 cm for a resolution in plane of 312 μm. Axial scout images were obtained, as well as quasi-coronal images perpendicular to orbital axis and quasi-sagittal images parallel to the orbital axis. Some imaging was performed after peripheral intravenous bolus administration of gadodiamide, 0.1 mmol/kg. Gadodiamide is a paramagnetic contrast agent commonly used in clinical MRI ²⁴ and for the first-pass MRI perfusion imaging of human EOMs. ²⁵  

Digital MR images were transferred to computer (Macintosh; Apple Computer, Cupertino, CA), converted to 8-bit tagged image file format (TIFF) using locally developed software, and quantified with NIH Image (available by ftp at zippy.nimh.nih.gov/ or at http://rsb.info.nih.gov/nih-image; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD). The location of the IO muscle in each image plane was defined by its area centroid, equivalent to the center of gravity of a shape of uniform density and thickness. The image plane containing the center of the IR muscle in central gaze was designated image plane zero, taking more nasal planes to be negative and temporal planes to be positive. This approach permitted averaging across subjects. 

Orbital Tissue Removal

Because dissection unavoidably alters the elastic suspensions of the orbital connective tissue network and may alter apparent anatomic relationships, whole human and monkey orbits were prepared intact for histologic examination. Five human orbits (ages, 44–93 years) were exenterated en bloc at autopsy by an intracranial approach within 24 hours of death. These were fixed for at least 5 days in 10% neutral buffered formalin (NBF) with periorbita intact but separated from bony support. Three additional human orbits (ages 17 months to 57 years) were obtained from a tissue bank (IIAM, Scranton, PA), in heads fresh frozen shortly after death. The heads were slowly thawed in NBF. One human orbit, stillborn at 30 weeks’ gestation, was obtained fresh at autopsy and fixed by immersion in NBF. Four monkeys, a rhesus (Macaca mulatta), a fascicularis (M. fascicularis), a cebus (Cebus apella), and a nemestrina (M. nemestrina) were killed in conformity with recommendations of the American Veterinary Medical Association and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and perfused with fixative through the aorta. The intact, fixed right orbits of the four humans and four monkeys were then removed in continuity with the eyelids and orbital bones, the latter being carefully thinned under magnification by using a high-speed drill. These fixed right orbits were then decalcified for 24 hours at room temperature in 0.003 M EDTA and 1.35N HCl. 

At autopsy of an adult male human conducted within 8 hours of death, the globe was enucleated using a surgical technique and magnification, and the IO was isolated and excised. Orientation of the excised IO was carefully maintained during fixation and later embedding for transverse sectioning. The opposite orbit was exenterated en bloc by an intracranial approach. After NBF fixation of the whole orbit, the region of the IO was carefully blocked without dissection in continuity with the adjacent sclera and connective tissues. This IO specimen was later embedded for sectioning in the quasi-sagittal plane corresponding to the MRI studies. 

Histologic Preparation

Human and monkey orbital tissues were examined microscopically in serial sections after staining for muscle and connective tissue constituents. Formalin-fixed tissues were embedded in paraffin and processed for sectioning at 10 μm thickness, as previously described. ^{1 2} Whole orbits were serially sectioned in the coronal plane. One isolated human IO was step sectioned transversely at intervals of 200 μm, whereas another was serially sectioned at 10-μm thickness in the coronal plane. Masson’s trichrome stain was used to show muscle and collagen, and Van Gieson’s stain to show elastin. ²⁶ As previously described, ² SM was confirmed using a monoclonal mouse antibody to human SM α-actin (Dako, Copenhagen, Denmark). 

Results

The IO muscle path was evaluated as a function of gaze in living humans by MRI. Imaging was performed without gadodiamide contrast in 15 orbits of 9 normal adult volunteers of mean ± SD age 55 ± 16 (range, 19–60) years and with contrast in 24 orbits of 12 volunteers of mean age 29 ± 11 years. Gaze angles used in imaging were determined directly from MRI images by measurement of the position of the globe–optic nerve junction. A typical data set of 2-mm-thick quasi-sagittal image planes, obtained without contrast, is illustrated in Figure 1 . The vertical white lines in the third and fourth columns of Figure 1 mark the anterior edge of the IO in supraduction. Note that the anterior IO edge moved posteriorly in infraduction, bringing it well posterior to the white lines. The uniform, horizontally oval IO cross sections in the right three columns of Figure 1 indicate that in this region, at and medial to the IR crossing, the IO followed a straight path nearly perpendicular to the quasi-sagittal plane. As illustrated in the left two columns of Figure 1 , the IO cross section lateral to the IR broadened and curved, indicating that it no longer followed a straight path. Lateral to the IR, the IO curved to insert on the globe, following a path varying with gaze direction. Based on the path inflection at the IO crossing of the IR, this general site was regarded as the location of the IO pulley, an interpretation consistent with histologic data below. 

For analysis of the IO’s path and cross section, sets of 17 contiguous 2-mm-thick quasi-sagittal image planes were obtained without contrast in central gaze, 24° infraduction with 5° abduction, and 24° supraduction with 13° adduction. Supraducted adduction was chosen to maximize IO contractile change, which preliminary experiments suggested were smaller in simple supraduction. The quasi-sagittal plane coordinates of the IO pulley as determined from IO centroids in image plane zero are plotted in Figure 2 for these gaze positions. There was a significant anterior, superior shift in supraduction, compared with a posterior, inferior shift in infraduction. At the point of crossing the IR center, the IO moved 4.3 ± 0.3 mm anteriorly from infraduction to supraduction, approximately 53% of the calculated travel of the IR insertion based on the gaze shift and the ocular radius. Lacking adequate anatomic references to standardize a coordinate system across subjects, we did not attempt to compute absolute coordinates of the IO pulley in central gaze. 

The relationship of the IR OL to the IO pulley was demonstrated by imaging without contrast in the secondary gaze positions of supraduction and infraduction. This is illustrated in Figure 3 , a quasi-sagittal image through the IR in which the typical feature of a bright intensity demarcation, presumably a fatty septum, is seen between the GL and OL. In supraduction, both the GL and OL were relatively small and the IO relatively anterior (Fig. 3) . In infraduction the OL, contracting more than the GL, can be seen to displace the IO posteriorly by dark, anteroposteriorly oriented connective tissue bands. 

Although the portion of the IO nasal to its pulley was straight and thus easily imaged in a quasi-coronal plane, the IO path temporal to its pulley curved superiorly and posteriorly along the globe surface to the scleral insertion in a gaze-direction–dependent path. Quasi-coronal plane MRI was explored in all secondary and tertiary gaze positions. Of these, supraducted adduction, which moved the IO insertion inferiorly and anteriorly, resulted in optimal imaging of the entire IO path in a single quasi-coronal plane. Simple adduction also produces a planar IO path, but supraducted adduction has the additional advantage of moving the LR and IR pulleys anteriorly so that these also lie in the plane of the IO path. When imaging in supraducted adduction was combined with gadodiamide contrast injection, which enhanced the EOMs but not their encircling pulleys, it was possible to demonstrate directly the consistent relationship of the IO to the IR and LR pulleys in 12 volunteers. Figure 4 is such a study of a representative left orbit, where supraducted adduction can be directly verified by the position in the lens in Figures 4D 4E 4F . The IO can be seen to follow a straight path from its origin on the bony nasal orbital floor to the region of the IR pulley, seen as a dark ring around the bright, enhancing IR muscle belly (Fig. 4F) . The connective tissue sheath of the IO can be seen as dark bands flanking the bright IO belly. The sharpest inflection in IO path was at the lateral aspect of the IR, at the insertion of the NFVB (Fig. 4C) . Temporal to the IR pulley, the IO path can be seen to curve superotemporally within the image plane along the globe surface into continuity with the LR pulley, seen as a dark ring encircling the bright, enhancing LR belly. Of importance, the IO and its sheath can be seen to be in direct continuity with the inferior margin of the LR pulley (Fig. 4D) . Although MRI does not have sufficient resolution to establish that this continuity represents actual insertion, this evidence is provided histologically later in the article. More posteriorly, as seen in Figure 4A , the pulleys are no longer present, leaving only the enhancing IR and LR bellies. The medial (MR) and superior rectus (SR) pulleys are not visible in Figure 4 , having shifted considerably posterior to the image planes shown in this gaze position of supraducted adduction. 

IO Muscle Size and Contractility

The IO cross-sectional area was determined in each image plane in which the IO appeared. These data are plotted in Figure 5 , where it may be seen that nasal IO cross sections were greater in adducted supraduction than in central gaze, which were in turn greater than in abducted infraduction. The IO cross section lateral to the IR was quite variable and was thus not quantitatively analyzed, because in this region, the IO curved sharply to become tangential to the imaging plane. Mean IO cross section in image plane zero was 0.13 ± 0.01 cm². The volume of the IO in central gaze was 0.27 ± 0.01 mL (mean ± SEM) but did not change consistently with gaze, because of the absence of gaze-related changes in cross section at the IO ends. 

Contractility of the IO was assessed by subtraction in each image plane of its cross section in infraduction from that in supraduction (Fig. 6) . The greatest contractile change was at the IR crossing, with lesser changes nasally near the origin. Contractile changes were not reliably indicated temporal to the mid-IR crossing, where gaze-related path changes and obliquity of the imaging plane reduced the reliability of IO cross-sectional measurements. 

Histochemistry and Immunohistochemistry

Correlations with in vivo MRI were provided by microscopy. Masson’s trichrome staining readily demonstrated the laminar architecture of the IO muscle in human and monkey specimens. The formalin-perfused monkey specimens had ideal preservation of cellular architecture. Of all the monkey species studied, the orbital anatomy of the nemestrina most closely resembled that of the human. Figure 7 is a quasi-coronal section from the left orbit of a nemestrina monkey comparable to the orientation of the human MRI scan in Figure 4 , illustrating many of the same structures visible in the human MRI. With the trichrome stain, the smaller and more dark-purple–stained OL fibers were located on the orbital aspect of the IO in this specimen, whereas the larger and more bright-red–stained GL fibers were on the global aspect. The OL was not uniformly on the orbital surface of the IO in all specimens, but often appeared, even in the same specimen, on the anterior or posterior IO surface in relationship to connective tissues. The OL did not completely encircle the GL in any specimen. In all specimens, however, the IO muscle was surrounded by a collagen sheath that was highly condensed near the IR crossing into a ring that constitutes its pulley (Fig. 7) . Higher magnification consistently demonstrated that the OL fibers of the IO inserted not only into the collagen of the IO pulley, but also broadly into the collagenous sleeve of the IO along much of its extent, frequently with a dense insertion on the inferior part of the lateral rectus (LR) pulley. Insertion of OL fibers into the IO pulley and sleeve was not only confirmed by their entry into dense collagen by Masson’s trichrome staining (Figs. 8A 8B) , but in adjacent sections stained with Van Gieson’s method by the presence of dark, black-stained elastin fibrils uniting the muscle fibers to the connective tissues (Figs. 8C 8D) . Van Gieson’s stain also demonstrated the presence of dense deposits of elastin in the IO pulley and many parts of the sleeve adjacent to the OL. The consistent presence of black-stained elastin fibrils joining OL fibers to the pulley indicates a stout mechanical bond. 

In all monkey specimens (Fig. 7 , top inset) and in the younger human specimens (Fig. 9) , it was possible to identify insertion of the IO OL into the LR pulley. In the human this is illustrated in Figures 9A and 9B , where Masson’s trichrome staining indicates multiple OL fibers of the IO extending deeply into the dense collagen of the global aspect of the LR pulley. The adjacent section stained by Van Gieson’s method demonstrates black-stained elastin fibrils in and around the IO muscle fibrils on the LR pulley (Figs. 9C 9D) . 

The connective tissue around the IO was better developed in younger human and monkey specimens than in older humans. This is demonstrated in Figure 10 , showing the dense connective tissue sleeve around the entire extent of the IO lateral to the IR in the younger two human subjects (Figs. 10A 10B) , but almost complete attenuation of the connective tissue sleeves in the 57- and 93-year-old human specimens, with preservation of only the IO pulley ring adjacent to the IR pulley (Figs. 10C 10D) . In the oldest two specimens there was fatty infiltration within the IO and IR pulley rings, evident from the white voids in the collagen (Figs. 10C 10D) . There was no evidence of insertion of the IO on the LR pulley in the older two specimens. All the specimens illustrated in Figure 10 show extensive motor nerve arborization within the IO and arising from the entry of the NFVB into the posterolateral aspect of the IR pulley. 

The conjoined pulley rings of the IO and IR were consistently in continuity with a band of dense collagen and dense elastin running nasally. Whereas subsidiary connective tissue bands ran toward the periorbita, the densest connective tissue became continuous with the inferior aspect of the medial rectus (MR) pulley, and an inferior part inserted on the posterior lacrimal crest. Immunohistochemistry for human SM α-actin demonstrated SM in this band and in the connective tissue junction between the IR and IO pulley rings. Smooth muscle was increasingly abundant in the nasal part of this band, becoming a dense band continuing over the nasal aspect of the MR pulley. This dense SM band has been described classically as Müller’s peribulbar muscle, ^{2 10} and more recently as the inframedial peribulbar SM. ²⁷ Another mainly collagenous connective tissue band ran from the conjoined IO and IR pulleys to insert on the bony posterior lacrimal crest, representing the nasal insertion of Lockwood’s ligament. 

The motor nerve to the IO was traced in Masson’s trichrome–stained serial sections anteriorly from the orbital apex in the 17-month-old human specimen. The oculomotor nerve branched into inferior and superior divisions quite deeply in the orbital apex, and just anterior to this the inferior division bifurcated again into a well-demarcated and more superior branch to the MR and a more inferior branch to the IO and IR. More anteriorly still, but nonetheless approximately 16 mm posterior to entry into the IO, this latter branch further bifurcated into a division to the IR and another to the IO. At this level, all the motor nerves had collagenous sheaths of similar thickness. Approximately 10 mm posterior to its entry into the posterior IO surface, the motor nerve sheath became continuous with, and of similar thickness to, the relatively thin collagenous sheath enveloping the IR muscle. The thickness of the IO nerve sheath remained comparable to that of the IR muscle sheath as far anteriorly as 5.5 mm from its entry point into the IO. Further anteriorly, the nerve divided into a manifold of approximately six fascicles and acquired a thicker collagenous investiture. This investiture thickened progressively as it merged with the dense collagen of the IO sheath and pulley at the posterolateral aspect of the IR pulley. 

Discussion

Fundamental Functional Anatomic Finding

Using high-resolution MRI, we characterized the location, size, and gaze-related contractile changes in the IO muscle. The IO follows a gaze-dependent yet straight path from its origin on the inferonasal orbital rim to the point at the inferolateral aspect of the IR muscle, but, lateral to that, the IO follows a gaze-dependent and curved path over the globe to its scleral insertion that does not generally remain in a single plane. Analysis of the IO path provides strong evidence for the presence of an IO pulley near the crossing of the IR muscle. 

Gaze-related inflections in rectus EOM paths demonstrated by MRI have been used to define the functional locations of the rectus pulleys. ¹⁷ Imaging of the IO path in living humans now also indicates an inflection in its path roughly at the lateral aspect of the IR crossing indicating an IO pulley. The IO thus pulls from its functional origin at the pulley to the scleral insertion, irrespective of the relative location of the anatomic origin on the orbital floor. The IO does not pull in the direction heretofore assumed based on the shortest path from bony origin to scleral insertion. Furthermore, the IO pulley is mobile. It moves posteriorly in infraduction, and anteriorly in supraduction, for a total travel in this study of approximately 4 mm, or about half of the travel of the IR insertion. The IO pulley is partly coupled to the actively moving IR pulley by elastic tissues suggested in quasi-sagittal MRI views such as Figure 3 . Coupling of the IR and IO pulleys was earlier suggested by the finding that the IR pulley moves nasally in supraduction and temporally in infraduction, consistent with contraction and relaxation of the IO. ³  

Studies of rectus pulley locations have relied on area centroids of EOM cross sections as reasonable surrogates of EOM force centroids for determination of EOM paths. This approach assumes homogeneity of EOMs within cross sections. That assumption is reasonable for the rectus EOMs, which have motor nerve entry zones remote from the pulleys. However, the IO’s motor nerve entry zone from the NFVB is in the region of the pulley, and the stiff nerves probably resists bending stress focally to sharpen the IO path inflection. Although the IO centroid data in this study reasonably estimate the change in IO pulley location with vertical gaze, they do not define the absolute location of the IO pulley with high precision. Correlations with the histologic evidence discussed in the next section suggest that the entry point of the NFVB into the IO muscle on the posterolateral surface of the IR pulley, easily demonstrable on quasi-coronal (Fig. 4) and occasionally on quasi-sagittal (Fig. 1) MRI, be considered the current best estimate of the absolute location of the IO pulley. As seen in Figure 1 , this site in central gaze is posterior to the globe’s center and thus is consistent with the plane of the rectus pulleys. ¹⁷  

Although the IO path is not always constrained to lie in a single plane, in adducted supraduction the segment of the IO temporal to its pulley becomes coplanar with the straight nasal portion. In this gaze position, corresponding to the clinical diagnostic position for the IO, contrast MRI demonstrates anatomic continuity between the IO muscle and the IR and LR pulleys. It had been believed that pulley connective tissues could not be directly imaged. The novel capability of directly imaging rectus pulley locations by contrast MRI is likely to be valuable in future investigations of pulley behavior. 

Contractile IO thickening in supraduction is greatest at the point of IR crossing. Quantitative measures of rectus ²³ and SO ^{28 29 30} cross sections have been shown to correlate with EOM size and contractile function. The present study extends these measures of functional anatomy to the IO, permitting its future use to validate hypotheses about IO function in pathologic states such as superior oblique palsy. ³¹  

Fundamental Histologic Finding

Microscopic examination of whole orbits of monkeys and humans provided structural correlates of the gaze-related changes in the IO, IR, and LR paths. The extensive pattern of insertion of the IO’s OL distinguishes it from the nonocular rotary EOMs such as the levator palpebrae superioris, and the accessory lateral rectus muscle of the monkey, ³² and the retractor bulbi muscle of nonprimate mammals such as the cat ³² and rat. ³³ These latter three EOMs have no OL, unlike the well-developed one of the IO. The IO in humans and monkeys is consistently endowed with an encircling ring of collagen, stiffened by elastin and SM, conjoined with the similarly constituted pulley of the IR, and located at the point where MRI demonstrated inflection in the IO path in living subjects. This ring comprises the IO pulley. The APH proposes that OLs of the rectus EOMs insert on and translate their respective pulleys anteroposteriorly, whereas the GLs travel through the pulleys to insert on and rotate the globe. In living humans, contrast MRI indicated continuity of the OL of the IO with the IO, LR, and, IR pulleys. Microscopic examination in humans and monkeys demonstrated that this corresponds to insertion of OL fibers of the IO onto these pulleys, extending the APH to the IO. Contraction of the OL would nasally translate the IO pulley, and relaxation would allow the IO pulley to move temporally under the elastic tension of the remainder of the IO and surrounding connective tissues. Contraction of the IO’s OL would also move the LR pulley inferiorly. In distinction to the rectus OLs, the IO’s OL also inserted in the younger specimens on the relatively thick connective tissue of its sheath temporal to the IR. 

Kinematic Implications

The insertion of the IO’s OL and NFVB on the conjoined IO and IR pulleys and the coupling of the IO to the LR pulley complete a mechanical network consistent with Listing’s half-angle ocular kinematics (Fig. 11) . The APH, now supported by considerable MRI evidence in humans, proposes that in kinematic primary position (Fig. 11A) , the four rectus pulleys lie in a plane paralleling Listing’s plane. ^{5 17} Because the IO is mechanically coupled to both the IR and LR pulleys, its curved lateral path, which defines the rotational axis it imposes on the globe, is nevertheless constrained by these two anchoring points to parallel the plane of the rectus pulleys. (The lateral IO path is not quite a great circle, because the plane does not pass through the globe center.) The IO inserts on the sclera near the LR pulley. The rotational axis produced by the IO is thus determined by the orientation of its terminal segment between the IR and LR pulleys. This axis is necessarily orthogonal the rotational axes produced by the rectus EOMs and by definition parallel to Listing’s primary position. The couplings of the IO to the IR and LR pulleys thus constrain the action of the IO in Listing’s primary position to pure torsion around the line of sight, eliminating interference with the kinematics of the rectus EOMs. 

The coordinated control postulate of the APH proposes that, in secondary gaze positions such as supraduction, Listing’s half-angle rule is satisfied if each rectus pulley is located distance D₁ posterior to the globe center, equal to the distance D₂ from globe center to the scleral insertion. ^{5 9} Thus, the LR rotational axis depicted in Figure 11B tilts posteriorly by half the angle of ocular elevation α, as long as α is a trigonometrically small angle typical of the oculomotor range. To then allow the tertiary positions of supraducted ad- and abduction to be reached from supraduction, the pulley of the contracting SR moves posteriorly, and that of the relaxing IR moves anteriorly by distance D₃, maintaining equivalent positions relative to the globe to enforce the half-angle rule. ⁵ These pulley shifts result from contraction of the SR’s OL, and relaxation of the IR’s OL, both of which insert on their respective pulleys. Note that because of the coupling of the IO pulley to both the actively mobile IR pulley and to the fixed bony orbit, the IO pulley moves anteriorly by distance D₃/2, half the distance traveled by the IR, corresponding to the MRI observations in this study (Fig. 2) . Sagittal MRIs such as those in Figure 3 suggest that elastic bands couple the IO pulley to the IR pulley and IR’s OL, with the bands stretching to enable the IO pulley to move only half as far as the IR pulley. The new IO rotational axis in supraduction thus tilts superiorly by angle α/2, keeping it perpendicular to the LR rotational axis. Although the IO rotational axis is no longer around the gaze line in supraduction, it remains purely orthogonal to the half-angle axis behavior required of rectus EOMs, and so the IO does not compromise their kinematics. Because the IO rotational axis is not along the line of sight in supraduction, in a Fick coordinate system it also has horizontal and vertical actions. 

The described behavior is due to the activity of the IR orbital layer in positioning the IO pulley. Activity in the IO’s OL also appears to control pulley positions, in this case those of the IR and LR on which the IO orbital layer inserts directly. This is consistent with prior MRI demonstration of inferior shift in the LR pulley and nasal shift of the IR pulley, in supraduction ^{3 17} and contraversive head tilt (Demer JL, et al. IOVS 2003;44:ARVO E-Abstract 2736). In both of these cases, the IO contracts. Coupling of the LR and SR pulleys by a dense fibromuscular band ^{10 20} would be expected to translate the SR pulley laterally with IO contraction. As directly observed by MRI in humans, translations of the IR, LR, and SR pulleys tend to keep them aligned with ocular torsion produced by the IO muscle during static head tilt relative to gravity (Demer JL, et al. IOVS 2003;44:ARVO E-Abstract 2736). These pulley translations are thus consistent with recordings from burst neurons in monkeys showing torsional shift of rectus pulleys transverse to the EOM axes in the direction of ocular counterroll induced by the static VOR. ³⁴ The IR pulley also translates nasally during convergence, ³⁶ probably under the influence of the IO’s OL, but perhaps also due to contraction of the inframedial orbital SM. ^{19 27}  

The present study suggests that although humans even in the 10th decade of life exhibit an IO pulley onto which the OL of the IO attaches, there may be age-related degeneration of attachments of the IO’s OL to other connective tissues. The OL appears in the oldest specimens to lose its insertion to the temporal IO sleeve and to the LR pulley. These connective tissue changes would be expected to compromise the kinematics described herein, perhaps leading to violations of Listing’s law in older humans. Data on Listing’s kinematics are not available on adults in this age range, but they are known to exhibit substantial age-related decrements in the range of ocular ductions ³⁶ and inferior shift in horizontal rectus pulley positions, ³⁷ perhaps due to connective tissue alterations. ¹⁹  

Clinical Implications

A stiff mechanical anchoring of the surgically transposed IO insertion is described at the NFVB as the motor nerve enters the posterior aspect of the IO lateral to the IR. ²⁰ The NFVB enters the IO within the dense connective tissue of the conjoined IO and IR pulleys, and its fascicles arborizing within the IO muscle may also locally augment its stiffness. The NFVB also coincides with the densest part of the insertion of the OL of the IO on its pulley. Surgical division of the NFVB would necessarily entail disruption of the OL insertion. It has been proposed that the NFVB’s connection to the orbital apex allows the NFVB to function as an ancillary IO origin after anterior transposition of the IO insertion to the side of the IR insertion. ²⁰ A problem with this concept is that, although the anterior portion of the NFVB is thick and heavily reinforced with collagen and elastin, the posterior portion of this lengthy structure is much thinner and contains sparse connective tissue. The effective stiffness of the NFVB should, like the weak link of a chain, be limited to that of its thinnest and most compliant region. The data herein support a modestly revised interpretation, that the sharp inflection in the path of the anteriorly transposed IO at the NFVB entry is not due to mechanical forces transmitted from the NFVB’s origin in the orbital apex, but rather to its colocation with the mechanically robust tissues of the IO pulley. It is also difficult to reconcile the concept of a long yet uniformly unyielding NFVB ²⁰ with the present observations of significant changes in IO pulley position with normal vertical gaze shifts. Figure 1 demonstrates that the path of the NFVB is not straight. The region of the NFVB entry into the IO coincides with and probably reinforces the IO pulley both normally and after surgical transposition of the IO’s scleral insertion. 

The IO connective tissue sleeve was considerably more extensive in children than in adults, and was markedly attenuated in the very elderly subject. This presumed degeneration is likely to be part of a generalized attenuation and dehiscence of adnexal connective tissues. The lower eyelid moves with the IO, presumably through coupling to its pulley, and age-related involution of these retractors causes involutional entropion. ²¹ Corresponding degeneration of the IO’s connective tissue attachments may be one factor underlying the recognized deterioration of supraduction in older people, ³⁷ and asymmetrical degeneration could cause small vertical heterophorias. Surgery on the IO typically involves exposure by incisions through inferotemporal conjunctiva where the IO is on the global side of Tenon’s fascia. The white connective tissue sleeve of the IO is dissected to expose the EOM fibers, disinserting the OL from the connective tissue sleeve. Surgical exposure of the IO would thus alter its mechanical action, even without further manipulation of its belly or scleral insertion. Variability in the outcome of IO surgery may be partly due to variations in the anatomy and surgical dissection of the IO sheath. Consideration of these issues may improve the predictability of IO surgery. 

 Supported by National Eye Institute Grant EY08313 and Core Grant EY00331 and an unrestricted award from Research to Prevent Blindness (JLD). JLD is the David and Laraine Gerber Professor of Ophthalmology.

 Submitted for publication February 12, 2003; revised April 8, 2003; accepted April 9, 2003.

 Disclosure: J.L. Demer, None; S.Y. Oh, None; R.A. Clark, None; V. Poukens, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Sets of contiguous, 2-mm-thick quasi-sagittal MRIs of a normal human right orbit, obtained in supraduction, central gaze, and infraduction. Vertical lines: anterior border of the IO muscle in supraduction, which can be seen to move posteriorly in infraduction. Note the contractile thickening of the IO in the middle column images, where the cross-sectional area is larger in supraduction than in infraduction. IR, inferior rectus muscle; LPS, levator palpebrae superioris muscle; LR, lateral rectus muscle; ON, optic nerve; SO, superior oblique muscle; MR, medial rectus muscle; NFVB, gray, L-shaped neurofibrovascular bundle entry to the IO.

Figure 2.

 

Mean quasi-sagittal plane coordinates of the human IO pulley in 15 orbits obtained from MRI in infraducted abduction and supraducted adduction. Note the anterior–superior shift in supraducted adduction, and the posterior–inferior shift in infraducted abduction.

Figure 3.

 

Quasi-sagittal MRI of human orbit showing action of the inferior rectus (IR) muscle on the IO pulley during vertical duction. The inferior rectus (IR) global layer (GL) thickened moderately in infraduction and inserted on the globe, whereas the orbital layer (OL) exhibited marked and focal contractile thickening as it inserted on the coupled IO and IR pulleys by dark elastic bands. LPS, levator palpebrae superioris muscle; ON, optic nerve; SR, superior rectus muscle.

Figure 4.

 

Gadodiamide enhanced, quasi-coronal MRIs of a left orbit in supraducted adduction. This gaze position shifted the temporal IO muscle path into the imaging plane to demonstrate the anatomic relationship to the inferior rectus (IR) and lateral rectus (LR) pulleys. Note in (D) the straight IO muscle path adjacent and inferior to the IR pulley and the curved IO muscle path temporal to the IR pulley running into continuity with the dark LR pulley (✶). The IO sheath is visible as a dark signal in (C) through (F). In (C) the IO sheath is also shown to be in continuity with the LR pulley. The LR pulley ring is most clearly visible at the right of the arrow in (C). Posterior to the pulley rings (A and B), only the enhancing LR and IR muscle bellies appear. In this gaze position of supraducted adduction, the medial and superior rectus pulleys have moved posterior to the imaging planes shown. LE, lateral enthesis suspending LR pulley. ME, medial enthesis suspending MR pulley.

Figure 5.

 

Mean cross-sectional area of 15 IO muscles in central (primary) gaze, supraducted adduction, and infraducted abduction.

Figure 6.

 

Mean contractility of 15 IO muscles showing the change in cross section from infraducted abduction to supraducted adduction.

Figure 7.

 

Coronal section of the left orbit of a nemestrina monkey stained with Masson’s trichrome showing the region of the IO and inferior rectus (IR) pulleys, and the anterior part of the lateral rectus (LR) pulley. Note the smaller and more darkly stained orbital layer (OL) fibers, and the larger and more bright-red–stained global layer (GL) fibers. The IO is surrounded by a thick sheath of dense, blue-stained collagen. Insets: OL fibers in continuity with the collagenous IO sleeve. Superiorly, this sleeve becomes continuous with the LR pulley, and inferiorly, with the dense collagenous ring that constitutes the IR pulley.

Figure 8.

 

Adjacent 10-μm-thick coronal sections of right orbit of a 17-month-old human showing the IO muscle just lateral to the inferior rectus (IR) muscle. The areas within the rectangles in (A) and (C) are magnified in (B) and (D), respectively. (A) Masson’s trichrome stain at low power shows the IO orbital layer (OL), now twisted toward the globe, in continuity with the dense blue collagen of the IO sleeve and IR pulley, which continues out of the field to the right. (B) Higher magnification shows light red OL fibers entering the dense blue collagen of the connective tissue. (C) Van Gieson’s elastin stain shows dense deposits of black-stained elastin fibrils within the connective tissue adjacent to the OL of the IO. (D) At higher magnification, black-stained elastin fibrils are abundant in the connective tissue, and some are in continuity with OL fibers.

Figure 9.

 

Adjacent 10-μm-thick coronal sections of the right orbit of a 4-year-old human showing the IO muscle just inferior to the lateral rectus (LR). The areas within the rectangles in (A) and (C) are magnified in (B) and (D), respectively. (A) Masson’s trichrome stain at low power shows the inferior part of the LR and part of the global region of its collagenous pulley ring stained blue. (B) At higher power, reddish purple individual fibers of the OL of the IO are embedded in the dense blue collagen of the LR pulley. (C). Van Gieson’s elastin stain at low power failed to resolve elastin in the region of the LR pulley. (D). At higher power, elastin fibrils are visible in and around the OL muscle fibers terminating on the LR pulley. Note dense black staining of the internal elastic lamina of the artery at bottom right.

Figure 10.

 

Masson’s trichrome stain of 10-μm-thick coronal sections of the right orbits of four humans of various ages. Note dense motor nerve arborization (purple) in the IO muscle in all specimens, because this region is near the entry of the NFVB. (A) In the 17-month-old specimen, note the thick blue collagen sleeve around the orbital and global surfaces of the IO muscle, as well as condensed collagen encircling the IO and the inferior rectus (IR) to form the IR pulley. (B) Similar thick collagen was seen in the 4-year-old specimen. (C). In the 57-year-old specimen, the collagenous sheath of the IO was markedly attenuated, whereas the blue-stained IO and IR pulley rings (left of IR) were preserved but had mild fatty infiltration (white). (D) In the 93-year-old specimen, there was virtually no collagenous sheath on the lateral part of the IO, and the IO pulley ring showed marked fatty infiltration (white).

Figure 11.

 

Lateral schematic view of IO and rectus pulleys. (A) Rotational axis of the lateral rectus (LR, vertical gray arrow) is perpendicular to the segment from its pulley to its insertion. Because the IO pulley is coupled to the LR and IR pulleys, the segment of IO between these pulleys lies parallel to LR rotational axis, so that the IO’s rotational axis (horizontal gray arrow) is parallel to the primary position. (B) Because the distance D₁ from LR pulley to globe center is equal to distance D₂ from globe center to LR insertion, supraduction to angle α causes the LR rotational axis (short gray arrow) to tilt by angle α/2, in accordance with Listing’s law. Relaxation of the IR orbital layer allows its elastic suspension to move the IR pulley anteriorly by distance D₃, and partial coupling of the IO pulley to the IR pulley moves the IO pulley anteriorly by half of this amount D₃/2. This pulley shift causes the terminal IO path to remain perpendicular to the anterior LR path, so that the IO rotational axis (long gray arrow) remains perpendicular to the LR axis.

The authors thank James Lynch and Lawrence Tychsen for generously providing some of the anatomic specimens and Nicolasa De Salles, Frank Henriquez, and Zita Jian for technical assistance.