Glenohumeral Stability

Biomechanical Properties of Passive and Active Stabilizers

Abstract

The shoulder is characterized foremost by its mobility and large range of motion. The glenohumeral joint is notable for its relative lack of bony constraint, relying heavily on the congruent articulating surfaces and surrounding soft tissue envelope for static and dynamic stability. Effective function in the articulation is achieved by a complex interaction between the various articular and soft tissue restraints. The rotator cuff muscles center the humeral head in the congruent glenoid fossa through the midrange of motion, when the capsuloligamentous structures are lax. However, incongruent joints, especially in positions of loading asymmetry (in external rotation), have larger translations that occur at the extremes of motion. Excessive translations are then effectively restricted by the mechanical properties of the inferior glenohumeral ligament. When the capsule is tightened anteriorly it results in an anterior tether and causes an associated posterior shift in contact on the glenoid. The posterior migration of the humeral head center and glenohumeral contact are again more pronounced in shoulders with reduced congruence. Additional studies of normal motion in different planes, the effects of rotator cuff pathology and dysfunction on the kinematics of the joint, proprioception of the capsule, and biomechanical tests of the inferior glenohumeral ligament and other components of the joint capsule at strain rates associated with injury, need to be conducted to understand the specifics of normal shoulder function and the pathophysiologic processes that occur during shoulder degeneration.

The concerted action of the active stabilizers (deltoid, biceps, and rotator cuff muscles) and the passive restraints (articular surfaces, osseous structures, and ligaments) is necessary for the purposeful function of the shoulder articulation. During the late 1970s, and continuing into the 1980s, there was an explosion in interest in the shoulder, and dramatic clinical advances were made in the treatment of pathologic conditions such as glenohumeral instability, rotator cuff disease, and shoulder arthritis. However, despite the fact that these treatments, especially surgical procedures, were explicitly designed to restore normal shoulder mechanics, little is known of normal glenohumeral and subacromial articular geometry, kinematics, and contact, and even less of the pathomechanics present in disease states or of the mechanical alterations produced by surgical reconstruction.

The humeral head has a surface area that is approximately 3 times that of the glenoid,^61,65 giving rise to a lack of bony restraint around the shoulder. The relative lack of bony restraint in the glenohumeral joint contributes to the shoulder possessing the largest range of motion (ROM) of any diarthrodial joint in the human body. In the absence of bony restraint, stability to the glenohumeral joint is provided by the articulating surfaces, capsular and ligamentous structures, and the synchronous activity of the rotator and deltoid muscle groups. Soslowsky and coworkers⁶⁵ demonstrated that the articulating surfaces of the normal glenohumeral joint are nearly spherical, and the 2 mating cartilage surfaces are very conforming. In the normal shoulder, Kelkar et al^36-38 demonstrated that the motion of the glenohumeral joint may be aptly described as a ball and socket articulation, with small translations of the humeral head center through the large range of shoulder motion. In this conformation, contact on the glenoid surface remains relatively uniform, whereas contact on the humeral head is focal and migrates over its surface during joint motion.^17,66

The glenohumeral ligaments serve as static stabilizers, preventing excessive translations of the humeral head, especially at the extremes of motion.^{6,7,41,42,50,72,73,75,78} Townley⁷² described the entire anterior capsule as a capsular mechanism that acts to prevent anterior dislocations. More recently, Turkel and coworkers⁷³ demonstrated that no single structure stabilizes the glenohumeral joint in all positions, and that the position and tightness of the anterior structures varies with abduction and external rotation of the arm. Turkel and associates, and other researchers also defined 3 anatomic and functional regions of the inferior glenohumeral ligament (Fig 1), and reported that the inferior glenohumeral ligament is the primary static restraint against anteroinferior subluxation, especially with the arm at the clinical position of anterior instability.^{6,7,44,50,53,73}

Initiation and subsequent progression of cartilage degeneration can occur because of alterations in either the intrinsic material properties of any tissue component or in the mechanics of a diarthrodial joint, as a result of high focal contact stresses.^11,26,45,57 Recent studies have accurately determined the intrinsic properties of articular cartilage and ligaments,^4,6,45,46 yet there exists no consensus as to what constitutes normal or abnormal glenohumeral joint mechanics. Some researchers think that the normal glenohumeral joint behaves as a (minimally) constrained ball and socket joint, allowing only small translations,^{8,27,36,38,55} yet others think that the joint is relatively unconstrained thus allowing large translations through the ROM.²⁵ Furthermore, researchers have separately analyzed either the geometry of the glenohumeral joint^30,40,61,65 or its motion characteristics,^19,25,31,55 and thus, little quantitative information exists relating the 3 dimensional kinematics of this joint to its articular geometry.

As investigators have begun to explore these areas, there remains a surprising lack of consensus as to even the most fundamental aspects of shoulder function. Major controversies include the shape of the glenoid and humeral head surfaces (elliptical or spherical),^32,59,62 the conformity of the glenohumeral articulation (matching humeral head and glenoid radii of curvature or mismatched, with the glenoid being flatter),^61,62 and the normal kinematics of the shoulder (ball and socket rotation or rotation combined with large translations during normal motion).^{25,28,37,38,55} These inconsistencies in description of structure and function result from experiments with technologic limitations especially in the earlier investigations,^19,31,55 and differences in experimental protocol (active versus passive manipulation) in some of the more recent studies.^25,35,37 Significantly, this lack of consensus impedes basic and clinical researchers in the study of the cause and progression of shoulder diseases, prevents orthopaedic surgeons from developing and refining treatment interventions aimed at improving shoulder mechanics, and fails to provide a firm foundation for the design of prosthetic replacements intended to replicate normal joint structure and function.

Some of the recent results of investigations on the structural and material properties, biochemical composition and histology of the inferior glenohumeral ligament, the articular and subchondral bone geometry, contact and kinematics of the glenohumeral joint, and the structure function relationships that are present in the shoulder are reviewed.

TENSILE PROPERTIES OF THE INFERIOR GLENOHUMERAL LIGAMENT

Material properties refer to intrinsic mechanical characteristics of the material itself, whereas structural properties reflect the mechanical behavior of the structure as a whole, including material and geometric contributions to the load deformation response.⁴⁵ This section describes investigations into the structural and material properties of the inferior glenohumeral ligament, the major anteroinferior static stabilizer of the glenohumeral joint.

Structural Properties

The methodology for these studies have been established and previously described in detail.^6,45,69 Twenty-four fresh frozen human cadaver specimens were used. Using previously defined anatomic landmarks, the 3 anatomic regions of each inferior glenohumeral ligament (the superior band, the anterior axillary pouch, and posterior axillary pouch) as outlined by Turkel and associates⁷³ were isolated as bone ligament bone specimens. The length of each ligament was measured at 3 points using a precision caliper, and averaged. Thickness and width were also measured at 3 equally spaced portions along the length of each specimen, and averaged. Thickness was determined using a custom designed electric conductivity sensing digital micrometer that was accurate to ±3 μm.¹ For the controlled strain rate tensile testing experiment, a servohydraulic testing machine (MTS, Minneapolis, MN) was used.⁶ Each bone ligament bone specimen was mounted in a custom designed apparatus that included a bath of physiologic saline solution containing enzyme inhibitors (Fig 2). Using a 6 ° freedom grip assembly, each bone ligament bone specimen was aligned parallel to the prevailing ligament fiber pattern and coaxial to the applied load. The testing protocol included a preconditioning procedure that applied a 1.5 N load to the specimens, followed by 5 minutes of stress relaxation.^22,75 At controlled elongation rates (8 specimens each at 0.004 mm/second, 0.04 mm/second, and 4 mm/second), the specimens were elongated to failure in uniaxial tension. Total bone ligament bone strain was measured from the grip to grip motion, and midsubstance strain was measured using gauge marks placed on the surface of the specimen (capsular side) perpendicular to the fibers using the video dimensional analyzer technique.^6,56,77

The entire nonlinear stress strain curve from the toe region to failure was modeled using an exponential stress strain law of the form σ = A(e^Bε - 1), where σ is the stress, ε is the strain, and A and B are material coefficients.^6,2 The derivative of this exponential stress strain law, also known as the tangent modulus, is given by dσ/dε = Bσ + C. This linear function of stress indicates the stiffening with increasing stress seen in collagenous tissues. The coefficient B represents the rate of change of the tangent modulus with respect to stress, and the product AB (=C) represents the tangent modulus of the tissue as the stress approaches 0. A least square nonlinear regression curve fitting procedure was used to obtain the material constants A and B.

Elongation rates in 2 ranges were defined (slow: 0.004 mm/second or approximately 0.01%/second and 0.04 mm/second or approximately 0.1%/second; and fast: 4 mm/second or approximately 10%/second). For the slow strain rates, significant differences in total specimen strain at failure for the three inferior glenohumeral ligament regions were observed. The anterior axillary pouch failed at a higher bone ligament bone strain than either the superior band (p < 0.001) or the posterior pouch (p < 0.001). Midsubstance ligament strains to failure paralleled the results for the bone ligament bone specimen strain measurements. Interestingly, average midsubstance failure strains represented only 35% to 45% of the bone ligament bone specimen strain at failure indicating that significant strain variations existed along the length of each region of the inferior glenohumeral ligament, and that considerable strain must exist near the inferior glenohumeral ligament bone insertion sites.

For the fast strain rate, the bone ligament bone strain at failure showed the same trend as at the slow strain rates, with the anterior axillary pouch failing at a higher bone ligament bone strain than either the superior band or posterior axillary pouch (p < 0.05). In addition, the tensile stress at failure was statistically different between regions. The failure stress in the superior band and anterior axillary pouch were significantly greater than the failure stress for the posterior pouch (p < 0.05). The results of the structural testing at the slow and fast rates are summarized in Tables 1-4. From these tables, it can be seen that the superior band of the inferior glenohumeral ligament specimen was found to be significantly stiffer statistically when stretched at a higher strain rate than at a lower rate.

Material Testing (Material Properties, Biochemistry, and Histology)

In a separate study to test material properties, 8 normal fresh frozen human cadaver shoulders with a mean age of 60 years were used. Again, 3 anatomic regions (superior band, anterior axillary pouch, and posterior axillary pouch) of the inferior glenohumeral ligament were used. These specimens were dissected from the surrounding tissue, isolated with their bony attachments, and dimensionally measured according to previously established methodologies.^60,70 From each bone ligament bone specimen of the superior band and anterior axillary pouch, tissue was obtained for biochemistry, histology, and biomechanical testing. The ligament region of the posterior axillary pouch was too thin to section for biomechanical testing, and was, therefore, analyzed only for biochemistry and histology. A 3- to 4-mm wide ligamentous strip from the anterior of each bone ligament bone specimen was removed for biochemical analysis. The posterior 7 to 8 mm of each bone ligament bone specimen from the superior band and anterior axillary pouch was resected, also without bone, for preparation into dumbbell shaped specimens for biomechanical testing. The remaining ligament tissue from each region, including the bony attachments, was fixed in 10% neutral buffered formalin for histologic evaluation.

Dumbbell shaped ligament specimens were used to determine the intrinsic mechanical properties of the ligament, separate from the insertion sites.^16,56,70 One-millimeter thick slices from the intracapsular surface of the superior band and anterior axillary pouch were cut using a freezing stage sledge microtome. From these thin slices of tissue, uniform dumbbell shaped specimens were cut parallel to the direction of the fiber orientation using a razor edged mold. Again, thickness measurements of the narrow, central gauge section were made using the authors' custom designed electric conductivity sensing digital micrometer.^1,6 A Bausch and Lomb stereo zoom microscope (Rochester, NY), adapted with an Olympus precision x-y translation stage (Melville, NY) and coupled to a Microcode digital micrometer (Newport Beach, CA) was used for width measurement.^6,60 The accuracy of the stereo zoom microscope is better than ±12.5 μm. The cross sectional area, required to calculate the engineering stress, was determined for each specimen from the product of the mean width and thickness. Strain was determined using gauge marks placed perpendicular to the fibers on the surface of the specimen using the video dimensional analyzer technique.^6,56,77

Each dumbbell specimen was mounted in a lax position, to avoid undesired preloading, in a custom designed apparatus containing a physiologic saline solution with enzyme inhibitors at room temperature. Uniaxial tensile testing was performed with a servohydraulic testing machine (MTS 858) using a load cell with a resolution of 0.001 N.¹⁶ The testing protocol included a preconditioning procedure that applied a 0.07 N load to the specimen and then allowed it to stress relax for 5 minutes.^6,21,74 Each specimen was then elongated to failure at a strain rate of 0.1% per second.

Biochemical composition was determined for each of the 3 regions using standard assays.^15,23 Dry weight and water content were determined, and samples were processed for proteoglycan content, total collagen content, and amount of mature hydroxypyridinium crosslinks. Histology from the bone ligament bone portions of each of the 3 regions from four shoulders was evaluated in the longitudinal plane using hematoxylin and eosin stain and polarized light, and Verhoeff's stain for elastin. Two additional shoulders, aged 76 and 83 years, were dissected into the 3 regions of the inferior glenohumeral ligament, and the entire bone ligament bone specimen from each region was evaluated for histologic characteristics.

There were no statistical differences for ultimate tensile stress (;tw_u), ultimate tensile strain (ε_u), or tensile modulus (E) between the 2 regions. However, the superior band showed a trend toward higher ultimate tensile stress (16.9 ± 7.9 MPa) and tensile modulus (130 ± 47.9 MPa) than the anterior axillary pouch (10.4 ± 4.2 MPa and 100.3 ± 37.1 MPa). The mean ultimate tensile strain for the 2 regions was 16.8 ± 4.6%. In all the tests, failure always occurred within the gauge section.

The proteoglycan content (sulfated glycosaminoglycan per dry weight) was higher in the superior band than in the anterior or posterior axillary pouch, and this was significant between the superior band (2.73 ± 0.7 mg/gm) and the posterior axillary pouch (1.49 ± 0.4 mg/gm). All other biochemical parameters were not statistically different, with the following overall mean values: water content, 80.9 ± 2.5%; collagen per dry weight, 800.4 ± 91.7 mg/gm (or 80%); hydroxypyridium crosslinks as mole per mole of collagen, 0.715 ± 0.13 mol/mol; and sulfated glycosaminoglycan per dry weight, 1.94 ± 0.8 mg/gm (or 0.2%).

The anatomic features, as described by O'Brien and associates,⁵⁰ included a collarlike attachment at the humeral insertion in 5 shoulders; the remaining 3 shoulders had a V shaped attachment. A posterior band was palpable, though not well visualized, in the posterior fourth of the posterior axillary pouch in 1 of the 8 specimens. However, there was no difference in the geometric proportions of the posterior band and the remainder of the posterior axillary pouch.

The histology was evaluated in the longitudinal plane, and included the ligament substance and adjacent insertion sites. Longitudinally organized fiber bundles were observed in the ligament portion and had a consistent crimping pattern for each of the 3 regions. These fiber bundles appeared more uniform in the midsubstance, and less uniformly oriented with greater fiber bundle interweaving near the insertion sites. The superior band seemed to have the most pronounced fiber bundle interweaving in the midsubstance, as well as the insertion sites, compared with the anterior or posterior axillary pouch. The anterior axillary pouch had increased crimping with the least amount of interwoven fiber bundles compared with the other 2 regions. Intermediate patterns for crimping and fiber bundle orientation were observed in the posterior axillary pouch. Elastin was found in each region.

NORMAL AND ALTERED KINEMATICS OF THE GLENOHUMERAL JOINT

There is a vast amount of literature implicating mechanical instability as a condition predisposing the glenohumeral joint to degenerative change.^{10,13,44,47,48,51,53,68} In a joint with abnormal mechanics resulting, for example, from damage to a component of the surrounding soft tissue envelope, it is often necessary to surgically intervene and stabilize the joint. Clinical procedures that tighten the anterior structures of the glenohumeral joint are routinely performed to prevent recurrent anterior instability. However, little quantitative information is available on: (1) the influence of the existing articular geometry on the results of these procedures; and (2) the effects these procedures have in modifying joint mechanics, and thus the relationship to glenohumeral joint arthritis reported after overly tight anterior repair. This section describes investigations conducted that used stereophotogrammetry to study the effects of articular congruence, external rotation and anterior capsular tightening on the kinematics, contact patterns, and relative stability of glenohumeral abduction.^{2,3,29,36-38,65,66} This section also illustrates the differences that result in kinematic analyses which use subchondral bone surface data rather than articular cartilage surface data.

The protocol used to elevate the humerus is a modification of the method previously described.^18,36,66 Nine fresh frozen human cadaveric shoulders (average age 50 years; range, 42-59 years) were dissected free of all soft tissue with the exception of the deltoid, rotator cuff tendons, glenohumeral joint capsule, subacromial bursa, and coracoacromial and coracohumeral ligaments. The humerus was elevated using 6 flexible cables simulating the 3 heads of the deltoid and 3 rotator cuff muscles (infraspinatus and teres minor combined). Active elevation was performed in 10 ° increments in 3 modes: (1) starting rotation (defined as that amount of external rotation allowing maximum passive elevation); (2) clinically defined neutral (corresponding to 0 ° external rotation); and (3) tightened (corresponding to 0 ° external rotation after a constant tension tightening of the anterior capsule had been performed). At each angle of elevation, a stereogram (pair of convergent photographs) of the 2 sets of optical targets rigidly attached to the humerus and glenoid was taken using the stereophotogrammetry system. Once the ranges of motion were completed, the joint was disarticulated, and a stereogram was also taken of the articular surfaces.³ The cartilage was dissolved and another stereogram was taken of each bone surface and its associated set of targets. Using a least squares optimization, spheres that best approximated the cartilage and bone surface data were fitted to the geometric surfaces. The fitted spheres were used to determine the joint congruence, and the coordinates of the humeral head and glenoid centers of curvature at different angles of elevation. Kinematic and contact analyses were performed using the techniques described previously.^2,36,38,66

The average radii of the humeral head and glenoid cartilage surfaces were 25.5 ± 1.5 mm and 27.2 ± 1.6 mm, respectively. In all but 2 of the joints tested, the difference in the radii (R_G - R_H) was less than 2.5 mm. There was only 1 joint in which the humeral head cartilage had a larger radius of curvature than the glenoid (R_H - R_G = 1.2 mm).

The average radii of curvature for the humeral head and glenoid bone surfaces were 25.2 ± 0.7 mm and 33.4 ± 3.4 mm, respectively. There was no statistical difference between the radii of curvature for corresponding humeral head cartilage and bone surfaces. However, the radii of curvature for glenoid bone surfaces were significantly (p < 0.005) larger than the corresponding glenoid cartilage surfaces. All glenoid bone surfaces had a larger radius of curvature than the matching humeral head bone surfaces.

Ranges of translation are defined as the differences between the maximum and minimum excursions of the center of the humeral head in the lateral-medial, posteroanterior (PA), and inferior-superior directions during abduction. All results are expressed as mean ± standard deviation. In starting rotation, the lateral-medial, PA, and inferior-superior ranges of translation were 0.9 ± 0.2 mm, 2.2 ± 1.1 mm and 2.9 ± 1.2 mm, respectively. In neutral rotation, the lateral-medial, PA, and inferior-superior ranges of translation were 0.7 ± 0.3 mm, 1.2 ± 0.6 mm and 2 ± 0.7 mm, respectively. In tightened rotation, the lateral-medial, PA, and inferior-superior ranges of translation were 0.7 ± 0.3 mm, 1.5 ± 0.6 mm and 2.2 ± 0.8 mm, respectively.

Of the 81 ranges of translation (9 specimens × 3 directions × 3 modes of elevation), there were only 4 cases (3 in inferior-superior and 1 in PA, all in starting rotation) in which a range of translation exceeded 4 mm. The 1 case in PA occurred in the only joint in which the glenoid cartilage surface had a smaller radius of curvature than the mating humeral head. Two of 3 cases in inferior-superior were in joints that had a glenoid cartilage radius of curvature that was more than 2.5 mm that of the articulating humeral head. The third case in inferior-superior occurred in a joint that was subluxated inferiorly at 0 elevation, and had 2.8 mm of translation between 30 ° to 180 °. There was high correlation (r = 0.7) between incongruence and inferior-superior translation, with incongruent joints having larger anteroinferior translations and correspondingly shifted contact patterns (Fig 3).

In lateral-medial, the starting rotation values showed a trend toward being larger than in neutral rotation (p = 0.06). However, in inferior-superior and PA, the ranges of translation in starting rotation were significantly higher (p < 0.005) than in neutral rotation, consequently leading to anteroinferiorly shifted contact patterns in incongruent joints. The ranges of translation in neutral rotation were less than 3 mm for all joints in all directions.

In lateral-medial and inferior-superior, there was no significant difference before and after the tightening of the anterior capsule. However, the center of the humeral head in tightened rotation was consistently posterior to its position in neutral rotation (Fig 4), consequently resulting in posteriorly shifted contact patterns on the glenoid. The posterior shift in contact was largest in 1 of 2 joints that had a glenoid cartilage radius of curvature that was 2.5 mm larger than its articulating humeral head (Fig 5). The ranges of translation in tightened rotation were less than 3 mm for all joints in all directions.

Humeral head bone surface data yielded consistently larger translations than those obtained using the cartilage surface data. For starting rotation, in inferior-superior, the translation of the center of the bone surface was significantly larger (p < 0.05) than that of the cartilage center. In lateral-medial and PA, although not significantly different, the ranges of translation from bone data tended to be larger than those from the cartilage surfaces. In neutral rotation, in all 3 directions, the bone data yielded significantly higher ranges of translation than the articulating cartilage surfaces (inferior-superior and PA: p < 0.05; lateral-medial: p < 0.005).

DISCUSSION

The role of the glenohumeral ligaments in providing static stability to the glenohumeral joint is widely accepted.^6,44,73,75 However, studies^6,54,70,71 on the inferior glenohumeral ligament, and other investigations on the superior glenohumeral ligament and coracohumeral ligament⁸ have shown that the glenohumeral ligaments do not have the strength characteristics of ligaments in the knee^12,49,76,77 when tested as bone ligament bone preparations, and thus must function in concert with the shoulder muscles, primarily the rotator cuff, to restrain the humeral head and facilitate normal function of the glenohumeral joint. Turkel et al⁷³ described how tension in the inferior glenohumeral ligament varied with arm position, and other authors have illustrated the importance of the inferior glenohumeral ligament as a static restraint of the shoulder in limiting pathologic anterior translation of the humeral head.^51,68,73 Some researchers have suggested that the superior glenohumeral ligament plays a minimal role in stability,^50,51 whereas the middle glenohumeral ligament has been considered as a secondary restraint to large inferior and anterior translations.

Investigators have also attempted to determine the effect of the negative intraarticular pressure that exists in shoulders with intact capsules.^24,33 Gibb and coworkers²² determined that the force required to sublux the humeral head out of the glenoid fossa was significantly less in shoulders with vented capsules. However, it is unlikely that intraarticular pressure has a significant role in stabilizing the joint in high load configurations of the shoulder.³⁹

To determine the efficacy of the various glenohumeral ligaments in passively restraining glenohumeral translation, several selective cutting studies have been performed^10,67,75 investigating the role of various portions of the capsule, coracoacromial arch, and coracohumeral ligament. Bowen et al¹⁰ and Warner and coworkers⁷⁵ reported that at lower elevations, the superior glenohumeral ligament and the anterior band of the inferior glenohumeral ligament served as restraints to large inferior translations, and that at higher elevations, the anterior and posterior bands of the inferior glenohumeral ligament limited inferior translations. Schwartz et al⁶⁴ concluded, in agreement with the findings of Turkel et al,⁷³ that the inferior glenohumeral ligament and inferior and posterior portions of the capsule were the primary passive stabilizers to anterior translation.

However, all these studies have been performed while excluding the action of the dynamic stabilizers (rotator cuff and deltoid muscles), and consequently in absence of a realistic compressive joint reaction force. Thus, although these studies have played a role in the understanding of static factors restraining shoulder translation, they are unable to provide the relative contribution of the passive and dynamic stabilizers to humeral head motion. Unfortunately, recent kinematic analyses of the glenohumeral joint,^14,35-38 which do incorporate the function of the rotator cuff and deltoid muscles, have not focused on the contribution of the passive stabilizers. Consequently, the relative contribution of the active and passive stabilizers to glenohumeral motion is still unclear. To address this issue, additional studies need to be performed to provide a better biomechanical basis for preoperative planning, prosthesis design, and joint modeling.

The present investigations found structural and mechanical differences among the 3 anatomic regions of the inferior glenohumeral ligament. The superior band, as well as the anterior axillary pouch, exhibited significant strain rate dependent viscoelastic behavior.⁶⁹ Proteoglycan content was higher in the superior band, with a decreasing proteoglycan content from the superior band, anteroinferiorly, to the posterior axillary pouch, posteroinferiorly. A higher proteoglycan content may represent a greater requirement for the superior band to behave in a viscoelastic manner, as proteoglycan content has been shown to influence this biomechanical property.⁶³ The finding that there is a significant viscoelastic behavior for the inferior glenohumeral ligament in tension further supports its functional role in maintaining the position of the humeral head when the arm is in positions associated with anterior instability.

The inferior glenohumeral ligament has a near elastic behavior in its central region within the ligament substance, whereas regions closer to the bony insertions exhibit a more viscoelastic behavior.^6,69,71 These properties of the inferior glenohumeral ligament may help to explain the varying response, or functional adaptability, of the inferior glenohumeral ligament in stabilizing the shoulder over a wide range of positions at different rates of loading, as well as the patterns of failure seen clinically, as in Bankart lesions (insertion site) and capsular stretching (ligament substance). This information will help one to anticipate the disease present in anterior instability. The surgical reconstructive procedures must be flexible enough to correct either pathologic lesion or both, as they may be present in combination in the same patient.

Tensile strain at failure was significantly higher in the anterior axillary pouch, and each of the 3 inferior glenohumeral ligament regions experienced a significant amount of strain before failing. The geometry and biomechanical properties of the inferior glenohumeral ligament seem to be well suited for its role as the primary static anterior stabilizer of the glenohumeral joint. When the arm is abducted and externally rotated, a sudden traumatic event may overwhelm the dynamic stabilizers.⁴³ It is under such conditions that the static restraints (inferior glenohumeral ligament) may be subject to large forces applied rapidly. Estimates of glenohumeral joint capsule strain rates for pitching a baseball are approximately 100% per second.⁵² In comparison, even the fast rate tested (approximately 10%/second) is relatively slow, and was chosen to allow for accurate data collection within the authors' system. More advanced high speed optical strain measurements techniques must be developed before physiologically high strain rates can be applied to ligament testing.

The glenohumeral articular surfaces have also been associated with providing static stability to the joint. Congruence, a measure of the joint conformity can be defined as the difference in the radii of the humeral head and the glenoid articulating surfaces. The closer this difference is to 0, the more congruent is the joint. The glenohumeral joint is often described as a minimally constrained ball and socket joint, though unlike the hip anatomy which more truly resembles a captured ball and socket joint, the glenohumeral joint surface geometry is considered less of a stabilizing factor because of the smaller surface area of the glenoid and the apparent shallowness of the glenoid.

Historically, technology has been the limiting factor in determining articular geometry and studying joint kinematics. Radiography was the standard technique used in early shoulder research to study the structure and function of the glenohumeral joint.^19,20,31,55 However, plain radiographs depict only the incongruent bony surfaces and not the conforming articulating cartilage surfaces of the glenohumeral joint, leading to the persistent misconception of glenohumeral incongruity. The data on articular and bone geometry presented by Soslowsky et al⁶⁵ demonstrate that the articulating cartilage surfaces are much more conforming than the underlying bone surfaces. Because the glenohumeral joint articulates as a ball and socket joint, it is necessary to know the location of the center of the articulating surfaces, not that of the bone surfaces beneath the articulating surfaces as determined from a plain radiograph. Indeed, more recently, Kelkar and coworkers³⁷ demonstrated that kinematic analyses using subchondral bone data significantly overestimate the actual translations of the humeral head center. Magnetic resonance imaging studies enabling visualization of the articular cartilage provide a much better appreciation of glenohumeral joint congruency, and may allow a more correct description of in vivo shoulder kinematics.

Range of translation of the humeral head center is a quantitative measure of the functional stability of the glenohumeral joint in each of the 3 principal anatomic directions. The present studies have shown that during active abduction of the humerus in the scapular plane using simulated muscle forces, there are minimal translations in all 3 anatomic directions.^36-38 However, Harryman and coworkers²⁵ reported that excessive translations occurred especially at the extremes of passive cross body motion. These reported differences must be viewed with respect to the following 2 factors: (1) vastly different motions are being compared (scapular plane abduction in this study and cross body motion in the study by Harryman et al; and (2) passive manipulation of the humerus results in significantly larger translations of the humeral head (especially at the extremes of motion) than those obtained using simulated muscle forces.^35-38 Although passive manipulation of the humerus is an important clinical diagnostic tool, to establish a baseline for normal shoulder function, it is essential to consider the motion of the articulation in an active mode rather than in a passive mode. Most recently, investigators have conducted experiments to study not only the active role of the muscles, but also to simulate this active role in a dynamic mode.^14,79 Because of the complexity of designing control systems to reproduce the muscle loading across the shoulder, these studies have used fixed ratios of loading between the muscles (typically based on their physiologic cross sectional areas). These approximations of relative muscle activity have led to some surprising results. For example, Wuelker et al⁷⁹ reported that in normal shoulders, 11 mm of superior translation of the humeral head center occurred during dynamic humeral abduction. In a separate study, Boardman et al⁹ reported that in their dynamic model of simulated rotator cuff disease, there was no significant difference between the kinematics of the shoulders before and after the cuff tears had been created. However, recent results of active models, and the authors' clinical experience suggest that functional, incomplete, and complete thickness tears of the rotator cuff do alter the kinematics of the shoulder joint (nonpublished data, Flatow EL, Raimondo RA, Kelkar R, et al: Active and passive restraints against superior humeral translation: The contributions of the rotator cuff, the biceps tendon, and the coracoacromial arch. Presented at the Twelfth Open Meeting of the American Shoulder and Elbow Surgeons, Atlanta, GA 1996). Clearly, further improvements in dynamic modeling of the shoulder musculature need to be made before results from these experiments are considered representative of normal shoulder function. Ideally, prostheses designed to replicate normal function should incorporate the characteristics of active glenohumeral motion while accommodating for the increased stiffness (and consequently less deformability) of the prosthetic components compared with that of articular cartilage.

The position of external rotation, starting rotation, is associated with higher ranges of translation than neutral rotation. Abduction in starting rotation increases the PA range of translation by nearly 100% over that in neutral rotation. Therefore, while starting rotation increases the range of motion (higher maximal elevation), it also represents a mode of elevation in which the humeral head has larger translations. In starting rotation, the posterior head of the deltoid is substantially less effective because of a reduction of its moment arm about the axis of rotation. The resulting asymmetric action of the deltoid provides less anterior stability thereby allowing a greater range of translation. Furthermore, the inferior glenohumeral ligament provides an anterior restraint only against large humeral head translations rather than the continuous restraint provided in the posterior direction by the externally rotated middle deltoid through the ROM. In neutral rotation, the deltoid functions more symmetrically, thereby maintaining the humeral head in a more centered position. This result is significant because it may explain why patients with recurrent anterior instability tend to compensate by elevating their arms in relatively less external rotation.

There was no difference in the inferior-superior and lateral-medial translations of the humeral head between the shoulders before and after tightening of the anterior capsule. However, in the anteriorly tightened shoulders, the center of the humeral head consistently translated posteriorly because of the anterior tether, with incongruent joints having larger translations.⁵ Articular contact patterns on the glenoid exhibited a similar posterior shift, and a reduction in contact area. These contact area reductions and posterior shifts in kinematics and contact location may help to explain the clinical development of arthritis after tight anterior shoulder instability repairs, because abnormal joint loading may lead to articular surface degeneration, especially on the posterior glenoid. These results of posterior translation also illustrate the balance that exists between the passive and dynamic stabilizers in maintaining humeral head position. In the unaltered shoulders in neutral rotation, the capsule remains relatively lax through the midROM and tightens only at the extremes of motion. In the tightened shoulders, the capsule provides a continual anterior restraint that causes the center of the humeral head to move posteriorly throughout the entire ROM. It is possible that any such unidirectional tightness, either primary or iatrogenic, could cause instability in the opposite direction, and further investigation is necessary to test this hypothesis.

The differences in cartilage and bone anatomies cause a significant difference in results obtained from kinematic analyses of bone surface and cartilage surface data. Kinematic analyses using bone surfaces yield erroneously large translation values compared with those obtained using the articular cartilage surfaces. This result is better understood when one considers that the center of rotation during abduction is closer to the center of curvature of the humeral head cartilage surface than to the center of curvature of the humeral head bone surface.³⁷ Thus, even a pure rotation about the cartilage center (0 translation of the cartilage center) would result in some translation of the center of the bone surface (Fig 6). The present studies have demonstrated that through the ROM there are very small translations of the humeral head,^36-38 which can be better appreciated by observing the proximity of the finite helical axes to the center of curvature of the humeral head (Fig 7). These results suggest that investigations using radiographs overestimate the incongruence in the glenohumeral joint and translations of the humeral head center, and hence radiographic analyses should be performed and interpreted with caution.

The shoulder is characterized foremost by its mobility and large ROM. It facilitates activities of the upper extremity in several planes for activities of daily living. The glenohumeral joint is notable for its relative lack of bony constraint, more than any other joint, relying heavily on the congruent articulating surfaces and surrounding soft tissue envelope for static and dynamic stability. Effective function in the articulation is achieved by a complex interaction between the various articular and soft tissue restraints. The rotator cuff muscles center the humeral head in the congruent glenoid fossa through the midROM, when the capsuloligamentous structures are lax. However, incongruent joints, especially in positions of loading asymmetry (in the starting rotation positions), have larger translations that occur at the extremes of motion. These translations are then effectively restricted by the mechanical properties of the inferior glenohumeral ligament. When the capsule is tightened anteriorly, it results in an anterior tether, and causes an associated posterior shift in contact on the glenoid. The posterior migration of the humeral head center and glenohumeral contact are again more pronounced in shoulders with reduced congruence. Additional studies of normal motion in different planes, the effects of rotator cuff disease and dysfunction on the kinematics of the joint, proprioception of the capsule, and biomechanical tests of the inferior glenohumeral ligament and other components of the joint capsule at strain rates associated with injury need to be conducted to understand the specifics of normal shoulder function and the pathophysiologic processes that occur during shoulder degeneration.

Fig 1: . Anatomic drawings of the 3 regions of the inferior glenohumeral ligament (IGHL) from intracapsular and extracapsular views. MGHL = middle glenohumeral ligament; SGHL = superior glenohumeral ligament.

Fig 2: . Schematic of tensile testing apparatus showing: a, load cell; b, 3 ° freedom ball joint; c, specimen holder; d, ligament specimen; e, video dimensional analyzer camera; f, testing chamber containing physiologic saline; and g, X-Y translation stage. The actuator of the MTS machine connected to the load cell provides the sixth degree of freedom (translation in the Z direction).

Fig 3a-b: . Glenoid contact maps for an incongruent joint at 90 ° arm elevation showing: a, centered contact in neutral and b, anteroinferiorly shifted contact in external rotation.

Fig 4: . Average position of the center of the humeral head relative to the glenoid in the unaltered and tightened modes, demonstrating a consistent posterior translation in the tightened shoulders.

Fig 5a-c: . Results for an incongruent joint demonstrating the position of the center of the humeral head relative to the glenoid and glenoid contact before and after anterior tightening. Note the large posterior translation after tightening in a, the relatively centered contact of the unaltered joint in b, and the large posterior shift in contact after tightening in c.

Fig 6: . Schematic depicting how radiographs that delineate subchondral bone can potentially overestimate the translations that occur in the joint. Here, a pure rotation (0 translation) about the cartilage center (gray dot) results in translations of the bone center (black dot).

Fig 7: . Figure depicting the finite helical axes for 30 ° increments and their proximity to the center of curvature of the humeral head cartilage surface, representing the ball and socket nature of the normal joint during active elevation in the scapular plane. Only 4 increments are shown for visual clarity. The humeral head and glenoid surfaces are shown at 90 ° elevation.