The Effect of a Glenoid Defect on Anteroinferior Stability of the Shoulder After Bankart Repair: A Cadaveric Study*

ITOI, EIJI M.D.†; LEE, SEOK-BEOM M.D.‡; BERGLUND, LAWRENCE J. B.S.§; BERGE, LINDA L. A.S.§; AN, KAI-NAN PH.D.§, ROCHESTER, MINNESOTA

Journal of Bone & Joint Surgery - American Volume:
Article
Abstract

Background: An osseous defect of the glenoid rim is sometimes caused by multiple recurrent dislocations of the shoulder. It is generally thought that a large defect should be treated with bone-grafting, but there is a lack of consensus with regard to how large a defect must be in order to necessitate this procedure. Some investigators have proposed that a defect must involve at least one-third of the glenoid surface in order to necessitate bone-grafting. However, it is difficult to determine (1) whether a defect involves one-third of the glenoid surface and (2) whether a defect of this size is critical to the stability of the shoulder after a Bankart repair. The purposes of the present study were (1) to create and quantify various sizes of osseous defects of the glenoid and (2) to determine the effect of such defects on the stability and motion of the shoulder after Bankart repair.

Methods: The glenoids from sixteen dried scapulae were photographed, and the images were scanned into a computer. The average shape of the glenoid was determined on the basis of the scans, and this information was used to design custom templates for the purpose of creating various sizes of osseous defects. Ten fresh-frozen cadaveric shoulders then were obtained from individuals who had been an average of seventy-nine years old at the time of death, and all muscles were removed to expose the joint capsule. With use of a custom multiaxis electromechanical testing machine with a six-degrees-of-freedom load-cell, the humeral head was translated ten millimeters in the anteroinferior direction with the arm in abduction and external rotation as well as in abduction and internal rotation. With a fifty-newton axial force constantly applied to the humerus in order to keep the humeral head centered in the glenoid fossa, the peak force that was needed to translate the humeral head a normalized distance was determined under eleven sequential conditions: (1) with the capsule intact, (2) after the creation of a simulated Bankart lesion, (3) after the capsule was repaired, (4) after the creation of an anteroinferior osseous defect with a width that was 9 percent of the glenoid length (average width, 2.8 millimeters), (5) after the capsule was repaired, (6) after the creation of an osseous defect with a width that was 21 percent of the glenoid length (average width, 6.8 millimeters), (7) after the capsule was repaired, (8) after the creation of an osseous defect with a width that was 34 percent of the glenoid length (average width, 10.8 millimeters), (9) after the capsule was repaired, (10) after the creation of an osseous defect with a width that was 46 percent of the glenoid length (average width, 14.8 millimeters), and (11) after the capsule was repaired.

Results: With the arm in abduction and external rotation, the stability of the shoulder after Bankart repair did not change significantly regardless of the size of the osseous defect (p = 0.106). With the arm in abduction and internal rotation, the stability decreased significantly as the size of the osseous defect increased (p < 0.0001): the translation force in shoulders in which the width of the osseous defect was at least 21 percent of the glenoid length (average width, 6.8 millimeters) was significantly smaller than the force in shoulders without an osseous defect. The range of external rotation in shoulders in which the width of the osseous defect was at least 21 percent of the glenoid length was significantly less than that in shoulders without a defect (p < 0.0001) because of the pretensioning of the capsule caused by closing the gap between the detached capsule and the glenoid rim. The average loss of external rotation was 25 degrees per centimeter of defect.

Conclusions: An osseous defect with a width that is at least 21 percent of the glenoid length may cause instability and limit the range of motion of the shoulder after Bankart repair.

Clinical Relevance: The results of the present study suggest that measures to restore the arc of glenoid concavity may be beneficial, in terms of both stability and motion, for patients who have a glenoid defect with a width that is at least 21 percent of the glenoid length.

Author Information

†Department of Orthopedic Surgery, Akita University School of Medicine, Hondo 1-1-1, Akita 010-8543, Japan.

‡Department of Orthopedic Surgery, Hallym University, Sacred Heart Hospital, 896 Pyungchon-dong, Donan-Ku, Anyang, Kyunggi-do 431-070, Korea.

§Biomechanics Laboratory, 128 Guggenheim Building, Mayo Clinic and Mayo Foundation, Rochester, Minnesota 55905.

Article Outline

The prevalence of fracture or erosion of the anteroinferior part of the glenoid rim among shoulders with recurrent anterior dislocation has been reported to range from 8 percent (eighteen of 226) to 73 percent (116 of 158)6,21,23,27,29. A fracture of the glenoid rim that results in a large anterior fragment is known to cause anterior instability of the shoulder2,8. Biomechanical studies have demonstrated an inverse relationship between the size of the glenoid defect and the stability of the shoulder: the larger the defect, the less stable the shoulder13. Those studies have demonstrated that the anterior stability of the shoulder is related to the shape of the glenoid in the absence of the tension of the capsuloligamentous structures.

The effect of a glenoid defect on the stability of the shoulder after the repair of a Bankart lesion (detachment of the capsuloligamentous structures from the glenoid) continues to be investigated. Rowe et al.27 apparently were the first to describe the relationship between the size of the glenoid defect and the outcome of Bankart repair. Those authors did not find any significant difference with regard to the outcome of Bankart repair among shoulders with various sizes of osseous defects involving one-sixth to one-third of the glenoid surface. On the basis of that report, some have suggested that a defect involving at least one-third of the glenoid surface may necessitate a bone graft16,17, whereas others have stated that bone-grafting is necessary for the treatment of a “large glenoid defect” but have avoided a quantitative description, probably because the size of the defect is difficult to assess4,20,24. Many clinicians believe that a large defect of the glenoid must be treated with bone-grafting when a Bankart procedure is performed4,16,17,20,24, but a description of the precise indications for bone-grafting is still lacking.

There are two major mechanisms of glenohumeral stability: glenoid concavity and ligamentous tension. The creation of various sizes of osseous defects changes the extent of glenoid concavity, and the creation and repair of a Bankart lesion changes the ligamentous tension. Because the effect of glenoid concavity on stability was studied previously10 and because our major interest was to determine the stability of the shoulder in the presence of various sizes of osseous defects after a Bankart repair, we hypothesized (1) that performing such a repair after partial glenoid excision would not change glenohumeral stability as long as the repaired capsular ligaments were under constant tension and (2) that performing such a repair after partial glenoid excision would cause obligatory capsular shortening, with each centimeter of shortening reducing the range of external rotation by 360/(π x humeral head diameter) degrees. The purposes of the present study were (1) to create and quantify various sizes of osseous defects of the anteroinferior part of the glenoid rim and (2) to determine the effect of such defects on the anteroinferior stability and rotational motion of the shoulder after Bankart repair.

Back to Top | Article Outline

Materials and Methods

Quantitative Assessment of Osseous Defects

Sixteen dried scapulae from individuals of unknown age and gender were obtained from the Department of Anatomy for the first part of this study. The glenoid was photographed, with the camera aimed at the articular surface and with a scale placed next to each specimen to indicate its relative size on the photographs. The image was scanned into a computer with use of a scanner (Scanjet ADF; Hewlett-Packard, Camas, Washington). The projected surface area of the glenoid was calculated with use of AutoCAD software (version 13; Autodesk, San Rafael, California), and the average shape of the glenoid was determined as follows. An outer fitting circle (circumcircle) that fit the superoinferior diameter of the glenoid was constructed for each specimen, and all of the circumcircles then were enlarged to the same size and superimposed to create a single image (Fig. 1-A). On the basis of this image, the average distance of the glenoid rim from the center of the circumcircle was calculated for a full 360 degrees and the average shape of the glenoid was constructed (Fig. 1-B). This average shape then was scaled to fit the average size of the glenoid as seen on the original scanned images. An anteroinferior osseous defect was then simulated on the computer by cutting off the anteroinferior part of the rim with a line that was inclined 45 degrees from the longitudinal axis of the glenoid, which extended through the supraglenoid and infraglenoid tubercles (Fig. 2). The osteotomy lines were drawn through points representing 12.5 percent (line 1), 25 percent (line 2), 37.5 percent (line 3), and 50 percent (line 4) of the glenoid length, which corresponded to the diameter of the circumcircle (Fig. 2, line A-B). The width of the osseous defect was equal to 9 percent of the length of the glenoid when the osteotomy was performed at line 1, 21 percent when it was performed at line 2, 34 percent when it was performed at line 3, and 46 percent when it was performed at line 4. On the basis of these simulated osteotomy lines, templates of four different sizes (small [83 percent of the average size of the glenoid], medium [100 percent], large [111 percent], and extra-large [134 percent]) were made, with four slits corresponding to the osteotomy lines (Fig. 3). In the present study, osteotomies 1, 2, 3, and 4 are defined as the osteotomies that were performed at lines 1, 2, 3, and 4, respectively.

Back to Top | Article Outline
Stability Study
Preparation of the Specimen

Ten fresh-frozen cadaveric shoulders were obtained from individuals who had been an average of seventy-nine years old (range, sixty-one to ninety-nine years old) at the time of death. Each specimen was thawed overnight at room temperature before preparation. All soft tissues superficial to the rotator cuff muscles were removed. The rotator cuff muscles were elevated from the scapula, and the tendinous portions were bluntly separated from the capsule, from medial to lateral, to a level that was approximately one centimeter lateral to the glenohumeral joint line. Elevation of the rotator cuff tendons was discontinued at this level because it was thought that additional dissection might damage the capsuloligamentous structures. The muscles of the arm and the periosteum were removed from the humeral shaft, which was then fixed in the center of an aluminum sleeve with use of polymethylmethacrylate. By this step of the dissection procedure, the contour of the glenoid rim was visible from outside of the joint capsule. The medial margin of the scapula was osteotomized along a line parallel to the glenoid surface (Fig. 4). Two small Kirschner wires were passed through the glenohumeral joint space, one parallel to a line connecting the anterior and posterior aspects of the glenoid rim and the other parallel to a line connecting the superior and inferior aspects of the glenoid rim. The scapula was potted in an acrylic box (nineteen by nine by five centimeters) with use of plaster of Paris in a position such that the two Kirschner wires were parallel to the floor.

Back to Top | Article Outline
Mounting of the Specimen

The prepared specimen was mounted on a custom multiaxis electromechanical testing machine (Avalon Technologies, Rochester, Minnesota) that allowed free movement in the x-y plane (Fig. 5). A six-degrees-of-freedom load-cell (model 45E15A-E24E; JR3, Woodland, California) was attached to the x-y table, onto which the acrylic box with the potted scapula was fixed. The x and y axes of the load-cell were rotated 45 degrees in relation to the axes of the x-y table so that the force against anteroinferior movement (the vector sum of force x and force y relative to the x-y table) was detected as force x′ relative to the load-cell (Fig. 6). The humeral shaft was held in a clamp to keep the glenohumeral joint in 60 degrees of elevation in the coronal plane (30 degrees posterior to the scapular plane). The clamp was attached to a sliding device that allowed superior and inferior movement of the humerus. This movement was necessary in order to allow the humeral head to climb the slope of the glenoid surface as the head was shifted anteroinferiorly. A linear-position transducer (TR-50; Novotechnik, Stuttgart, Germany) was attached to the sliding device to measure the lateral movement of the humeral head along the z axis (that is, perpendicular to the glenoid surface). The transducer was calibrated against a linear encoder (Futaba Pulsescale; Futuba, Mobara, Japan) with digital readout mounted on a vertical milling machine (Bridgeport; Bridgeport Machines, Bridgeport, Connecticut). The calibration factor was calculated by fitting a line with least-squares regression. According to the information supplied by the manufacturer, the linearity of this transducer was 0.075 percent. A fifty-newton axial force was constantly applied to the humerus in order to keep the humeral head centered in the glenoid fossa. In the present study, direction is expressed according to the orientation of the glenoid: the x axis corresponds with the anterior-posterior direction; the y axis, with the superior-inferior direction; and the z axis, with the medial-lateral direction.

Back to Top | Article Outline
Testing Protocol

The x-y table was positioned so that the relative movement of the humeral head against the glenoid was in the anterior-inferior direction (inclined 45 degrees in relation to the x and y axes). First, the reference position was determined by measuring the lateral displacement of the humeral head along the z axis during a 6.0-millimeter translation in the superior-inferior and anterior-posterior directions. In both directions, the most medial position of the humeral head on the glenoid surface (the position where the center of the head was closest to the glenoid) was defined as the reference neutral position that was used for the subsequent displacement-control study28. Then, the humeral head was translated in the anteroinferior direction from zero millimeters (the neutral reference position) to 10.0 millimeters, was reversed to -4.0 millimeters, and finally was returned to zero millimeters. The rate of displacement throughout the cycle was 0.5 millimeter per second. Data regarding displacement and force along the x, y, and z axes were sampled at a rate of four hertz and were recorded with a personal computer. The displacement to -4.0 millimeters was performed to ensure a consistent hysteresis effect through the zero to 10.0-millimeter region of interest. The slow displacement rate of 0.5 millimeter per second was chosen to minimize viscoelastic effects14.

Although all specimens were tested with use of this ten-millimeter-displacement protocol, the analysis of force was based on a normalized displacement, which was proportional to the size of the glenoid. The displacement length of ten millimeters was used for the glenoid with the largest length (forty-one millimeters), and the displacement lengths for all other glenoids were downsized accordingly. Then, the peak force that was necessary to translate the humeral head the normalized distance (not the force that was necessary to hold the humeral head in the translated position) was determined for each test condition (to be described) (Fig. 7).

Back to Top | Article Outline
Test Positions

Two positions were chosen for testing: (1) abduction and external rotation and (2) abduction and internal rotation. The glenohumeral joint was elevated 60 degrees in the coronal plane in order to simulate a position with the arm in 90 degrees of abduction relative to the trunk25. External rotation was defined as the position 10 degrees internal from full external rotation (manually torqued), and internal rotation was defined as the position 10 degrees external from full internal rotation14. The angles were measured with use of a goniometer that was attached to the clamp. The angles of full external and internal rotation were recorded for analysis. The orientation of the humeral shaft was determined on the basis of the location of the bicipital groove, which is anterior with the arm in 10 degrees of internal rotation at the side15,20.

Back to Top | Article Outline
Test Conditions

Testing was performed under eleven sequential conditions: (1) with the capsule intact, (2) after the creation of a simulated Bankart lesion, (3) after the capsule was repaired, (4) after osteotomy 1, (5) after the capsule was repaired after osteotomy 1, (6) after osteotomy 2, (7) after the capsule was repaired after osteotomy 2, (8) after osteotomy 3, (9) after the capsule was repaired after osteotomy 3, (10) after osteotomy 4, and (11) after the capsule was repaired after osteotomy 4. A Bankart lesion was created by elevating the capsulolabral attachment subperiosteally from the glenoid from two o'clock to eight o'clock in the right shoulder and from four o'clock to ten o'clock in the left shoulder. Furthermore, the continuation of the labrum after detachment was disrupted at three o'clock in the right shoulder and at nine o'clock in the left shoulder. The Bankart lesion was repaired with use of three transosseous sutures made of braided polyester (number-2 Ethibond; Ethicon, Johnson and Johnson, Somerville, New Jersey). The type of Bankart repair that was performed in the present study involved placing the transosseous sutures at the lip of the glenoid in order to reestablish the glenoid concavity (as opposed to the type of repair in which the detached labrum is repaired to the glenoid neck, which does not reestablish the glenoid concavity). The repair was performed with the arm in 0 degrees of abduction and 30 degrees of external rotation relative to the scapular plane, which was equivalent to adduction and neutral rotation relative to the trunk19. The osseous defects were created with use of the osteotomy template that most closely fit the size of the glenoid. First, osteotomy lines 1 through 4 were marked on the glenoid surface. Next, the bone was resected at osteotomy line 1 with use of a reciprocating saw (AMSCO/Hall Surgical, Santa Barbara, California). The osteotomy was performed as perpendicular to the glenoid surface as possible. Larger osseous defects were created similarly, in a stepwise fashion, at osteotomy lines 2, 3, and 4. The displacement test was performed under each condition. We also manually checked the gross stability of the shoulder (subluxation and dislocation) with use of a load and shift test that was performed with the arm in adduction and neutral rotation as well as in abduction and neutral rotation without the application of the fifty-newton axial force.

To verify that the osseous resection had been performed as planned, the length of the glenoid remnant (Fig. 2, line A-C) was measured after the experiment. In addition, the diameter of the humeral head was measured with use of a caliper.

Back to Top | Article Outline
Analysis of the Data

The data were categorized into two groups: one in which the Bankart lesion was repaired and one in which it was not. As the purpose of this study was to determine the effect of an osseous defect on the stability of the shoulder, the data within each group were analyzed with use of repeated-measures analysis of variance, with the osseous defect as a within-group factor. When a significant effect was observed, the Newman-Keuls multiple-comparisons test was used to identify where the significance existed. The ranges of external and internal rotation in the group with Bankart repair were analyzed with use of the same method. The level of significance was set at p < 0.05. With ten specimens, the study had an 80 percent power to detect a difference in averages between any two of the experimental conditions that was equal to 1.0 standard deviation.

Back to Top | Article Outline

Results

Size of the Osseous Defect

The size of the osseous defect was determined after each osteotomy (Table I). To verify that the osseous resection had been performed as planned, the length of the glenoid remnant was measured after the experiment. The average length (and standard deviation) of the glenoid remnant after osteotomy 4 was 17.2 ± 2.6 millimeters. As the average glenoid length was 32.9 ± 5.0 millimeters, the length of osteotomy 4 (Fig. 2, line C-B) was planned to be an average of 16.4 ± 2.5 millimeters. The actual length was an average of 15.7 ± 3.0 millimeters, which was 96 percent of the planned length.

Back to Top | Article Outline
Stability in Abduction and External Rotation (Fig. 8)

With the shoulder intact, the average force that was needed to move the humeral head the normalized distance (average, 7.8 ± 1.1 millimeters) was 185 newtons. After the creation of the Bankart lesion, this force decreased to thirty-eight newtons. After the Bankart lesion was repaired, the force returned to only 108 newtons (58 percent of that in the intact shoulder). This was because force was tested immediately after the repair and therefore the healing process was not taken into consideration. Thus, the condition after the creation of the Bankart lesion and the condition after the repair of the Bankart lesion were used as the baseline conditions for the groups with and without repair, respectively. In the group without Bankart repair, stability (defined as the force that was needed to move the humeral head the normalized distance) decreased as the size of the osseous defect increased, with the forces that were measured after osteotomies 3 and 4 being significantly smaller than the baseline force (p = 0.020). Conversely, in the group with Bankart repair, stability did not decrease significantly, regardless of the size of the osseous defect (p = 0.106).

Back to Top | Article Outline
Stability in Abduction and Internal Rotation (Fig. 9)

The effect of an osseous defect on anteroinferior stability was more prominent with the arm in internal rotation. In both groups, stability decreased as the size of the osseous defect increased. In the group without Bankart repair, the forces that were measured after all four osteotomies were significantly smaller than the force that was measured in the absence of an osseous defect (p < 0.0001). In the group with Bankart repair, the forces that were measured after osteotomies 2, 3, and 4 were significantly smaller than the force that was measured in the absence of an osseous defect (p < 0.0001).

Back to Top | Article Outline
Gross Stability

With the arm in adduction and neutral rotation, the shoulders were unstable without the intra-articular pressure of the glenohumeral joint. In all ten shoulders, the humeral head was easily dislocated anteriorly with use of the manual load and shift test. With the arm in abduction and neutral rotation, the shoulders were stable anteriorly after Bankart repair in the absence of an osseous defect. However, two of the ten shoulders subluxated anteriorly after osteotomy 1, five shoulders subluxated after osteotomy 2, and all ten shoulders were grossly unstable (eight shoulders dislocated and two subluxated) after osteotomy 3.

Back to Top | Article Outline
Range of Motion (Fig. 10)

After Bankart repair, the range of external rotation decreased as the size of the osseous defect increased, with the ranges that were measured after osteotomies 2, 3, and 4 being significantly smaller than the range that was measured in the absence of an osseous defect (p < 0.0001). Conversely, the range of internal rotation was not affected by the size of the defect (p = 0.0529). As the average limitation in external rotation after osteotomy 4 (average width, 14.8 millimeters) was 37 degrees, the average limitation in external rotation was calculated to be 25 degrees for each centimeter of width of the osseous defect.

Back to Top | Article Outline

Discussion

Without Bankart repair, the stability of the shoulder progressively decreased as the size of the osseous defect increased. This finding is consistent with the results of previous studies that have demonstrated the importance of the glenoid concavity to the stability of the shoulder in the absence of the soft tissues10,11. However, as we hypothesized, the anteroinferior stability of the shoulder after Bankart repair remained unchanged, regardless of the size of the osseous defect, when the arm was in abduction and external rotation. In that position, the repaired capsuloligamentous structures between the glenoid rim and the articular edge of the humeral head became tight and seemed to prevent anteroinferior translation of the humeral head regardless of the length of the glenoid remnant (Figs. 11-A and 11-B). Similarly, Novotny et al.22, in a study of cadaveric shoulders, reported that the anterior stability of the shoulder with the arm in abduction and external rotation did not change in association with the change in the length of the capsuloligamentous structures after they had been repaired back to the glenoid rim. These findings indicate that the Bankart repair obviates the effect of glenoid deficiency when the arm is in abduction and external rotation, where the anterior part of the capsule is tight. Conversely, when the arm was in abduction and internal rotation, the anteroinferior stability of the shoulder decreased as the size of the osseous defect increased. In this position, the anterior part of the capsule was lax and the posterior part of the capsule was tight. When the glenoid rim was intact, the tight posterior part of the capsule seemed to prevent anteroinferior translation of the humeral head (Fig. 12-A). When there was a defect of the anteroinferior part of the glenoid, however, the humeral head shifted anteroinferiorly because the lax anterior part of the capsule did not obviate the effect of the glenoid deficiency (Fig. 12-B). It is apparent that the ligaments stabilize the glenohumeral joint only when they are under tension. In fact, qualitative assessment revealed that in the mid-range of motion (abduction and neutral rotation), where there was little effect of ligamentous tension, instability (dislocation or subluxation) increased as the size of the glenoid defect increased, with half of the shoulders becoming unstable after osteotomy 2 and all becoming unstable after osteotomy 3. Thus, in the present study, the stability of the shoulder after Bankart repair depended on the position of the arm and deteriorated in the presence of an osseous defect with a width that was at least 21 percent of the glenoid length (average width, 6.8 millimeters). It is possible that similar findings would be observed in cases of glenoid fracture. If a fracture fragment of this size were removed, the stability of the shoulder would be substantially reduced.

The range of motion also was affected by the size of the osseous defect. The range of external rotation was restricted when the Bankart lesion was repaired in the presence of an osseous defect that was as large as or larger than that associated with osteotomy 2. This was because repairing the capsule back to the glenoid rim in the presence of an osseous defect was equivalent to tensioning the capsule by closing the gap created by the defect. Lusardi et al.13, in a clinical study of twenty shoulders (nineteen patients), noted that anterior capsulorrhaphy may result in severe limitation of external rotation. The relationship between the amount of imbrication of the capsule and the range of motion also has been investigated in cadaveric studies. Black et al.3, in a study of six cadaveric specimens, demonstrated that two millimeters of imbrication of the anterior part of the capsule caused a 40 percent decrease in external rotation and that seven millimeters of imbrication caused a 67 percent decrease. On the basis of those results, we calculated that the average limitation in external rotation was 25 degrees per centimeter of imbrication. Matsen et al.17 reported that shortening the anterior part of the capsule and the subscapularis tendon by one centimeter limited external rotation of the humerus by approximately 20 degrees. We hypothesized, on the basis of theoretical calculations, that one centimeter of shortening would limit external rotation by 360/(π x humeral head diameter) degrees. As the average diameter of the humeral head was 44.1 ± 4.4 millimeters, we calculated that one centimeter of shortening would limit external rotation by 26 degrees. In fact, the measured limitation was 25 degrees per centimeter of defect. Thus, our hypothesis was proved.

As Gill et al.4 suggested, a balance between gaining stability and minimizing loss of motion is important. Occult anterior instability, which is thought to cause posterosuperior impingement7,12,34, may not be apparent unless a patient is involved in throwing activities. Montgomery and Jobe18 reported that a 15-degree limitation of external rotation was not a problem for a professional golfer, whereas a 5-degree limitation was disastrous for a professional baseball pitcher because of decreased velocity and performance. Thus, the need for bone-grafting should be decided on a case-by-case basis when an osseous defect is smaller than that associated with osteotomy 2. In throwing athletes who have an osseous defect, any method that restores the effective length of the anterior arc, either through bone-grafting or through elongation of the capsuloligamentous structures, appears to be justified in order to avoid a limitation of external rotation.

There have been many reports on the clinical results of the Bankart procedure. The presence of a glenoid fracture is known to cause chronic anterior instability2,8. Aston and Gregory2 reported the case of a patient who had recurrent anterior dislocations after sustaining a fracture that resulted in an anterior glenoid fragment involving slightly less than one-fourth of the articular surface. The shoulder did not redislocate after the fragment was fixed. Rowe et al.27 apparently were the first to describe the relationship between the size of the glenoid defect and the outcome of Bankart repair. Those authors did not find any significant difference among shoulders with various sizes of osseous defects involving one-sixth to one-third of the glenoid surface. Similarly, Protzman26 observed no relationship between the presence of a glenoid lesion and the outcome of a modified Bankart procedure. Thomas and Matsen30 and Wirth et al.35 performed a Bankart procedure without bone-grafting for all shoulders that had traumatic recurrent anterior dislocations with or without an osseous defect. The rates of recurrence were 2.6 percent (one of thirty-nine) and 0.9 percent (one of 108), respectively. On the basis of those reports, it appears that bone-grafting may not be necessary when an osseous defect involves less than one-third of the glenoid surface16,17,27.

In interpreting those reports, the greatest problem is that it is not clear how the investigators determined that the defect involved one-third of the glenoid surface. To the best of our knowledge, the only study to have included a quantitative assessment of the osseous defect of the glenoid was reported by Ungersböck et al.32, who evaluated the relationship between the width of the glenoid defect and the outcome of Bankart repair. In that study, none of twenty-six shoulders in which the osseous defect was less than three millimeters wide had a recurrence after the operation compared with one of three shoulders in which the osseous defect was at least three millimeters wide. Although the number of shoulders was small, the results of that study suggest the importance of the anteroinferior part of the glenoid rim to the stability of the shoulder after Bankart repair. The three-millimeter width cited by Ungersböck et al.32 is almost equivalent to the 2.8-millimeter width of the defect produced by osteotomy 1 in the present study. Although those authors reported one redislocation in a shoulder in which the defect was at least three millimeters wide, they did not specify the exact size of the defect in that particular case. In the present study, we expressed the size of the osseous defect as a percentage of the length of the glenoid. For the purpose of comparing our results with those in the literature, we also calculated the size of the defect as a percentage of the area of the entire glenoid surface (Table I). According to this conventional method, osteotomy 2, on which the stability of the repaired shoulder hinged, involved approximately one-fifth of the glenoid surface. To our surprise, however, it looked as if the defect produced by osteotomy 2 involved almost one-third of the glenoid surface. What Rowe et al.27 and others16,17 have described as one-third of the glenoid surface actually may be one-fifth. Thus, an area ratio, which has been commonly used, is not an accurate way of expressing the size of the defect because the determination of area is far more complex than the measurement of length. We believe that it is more accurate to express the size of a defect in terms of width (for example, 6.8 millimeters) or as a percentage of glenoid length (for example, 21 percent) rather than as an area ratio (for example, one-fifth).

The present study had several limitations. First, a simulated Bankart lesion is different from a Bankart lesion in vivo. Previous investigators have observed that a simulated Bankart lesion alone does not cause anterior dislocation1,28. They speculated that other factors, such as tearing of the posterior part of the capsule and functional deficiency of the rotator cuff muscles, also were responsible for anterior dislocation. In the present study, a simulated Bankart lesion was created not only by elevating the labrum from the glenoid but also by disrupting the continuity of the labrum. In our pilot study, we noticed that a shoulder did not become unstable even after the anteroinferior aspect of the labrum was totally elevated from the glenoid if the continuity of the labrum was preserved. As the humeral head passes through the space between the glenoid and the detached labrum during in vivo dislocation, the labrum must become either extremely elongated or disrupted. A review of the methods used by previous investigators1,28 to create a Bankart lesion suggests that the continuity of the labrum is one of the factors that may have affected the results. Elongation of the capsule in shoulders with recurrent anterior dislocation5,31,33 could be another factor that explains the discrepancy between the experimental and clinical settings.

Second, the orientation of the osseous defect in the present study may have been more inferiorly based than is the case in the clinical setting. The shoulder typically dislocates in the anteroinferior direction. However, as the glenoid is tilted anteriorly at approximately 20 degrees9, dislocation actually occurs in a slightly more anterior, rather than anteroinferior, direction in relation to the glenoid. We believed that this discrepancy would have little effect on the results as long as the osseous defect and the translation force were oriented in the same direction.

Third, we measured the range of rotation by applying the torque manually. This may have led to some error in the measurement of the range of motion. We tried to minimize the error by having the same individual (E. I.) apply the torque. The similarity between the range of motion that was measured during the experiment and the range that was predicted on the basis of our theoretical calculations indirectly proves the validity of the experiment.

Fourth, we limited the range of displacement to ten millimeters in order to avoid overloading the specimens. This range of displacement did not cause dislocation or subluxation. However, as the abducted and externally rotated shoulder was very stable in the anteroinferior direction when the capsulolabral structures were intact, the amount of increase in anteroinferior displacement observed in shoulders that had a defect with a width that was 21 percent of the glenoid length (average width, 6.8 millimeters) would be likely, in the clinical situation, to cause apprehension that the shoulder will subluxate anteriorly.

Finally, two test positions were chosen: (1) abduction and external rotation and (2) abduction and internal rotation. Some may argue that abduction and internal rotation is not the position of anterior instability. Others may argue that abduction and neutral rotation should have been chosen to test the effect of glenoid concavity. We had planned to perform the displacement test with the arm in neutral rotation as well as in external and internal rotation. However, in neutral rotation, the head was too unstable to be tested. For example, after osteotomy 4, the head was on the verge of the articular surface of the glenoid and it dislocated as soon as it was moved anteroinferiorly because there was no effect of ligamentous tension. Because of the fifty-newton axial load, the dislocated head could not be reduced during backward movement and, thus, the test had to be discontinued. As our major interest was the effect of ligamentous tension in shoulders with and without an osseous defect, we decided not to perform the displacement test with the arm in neutral rotation. Instead, gross stability with the arm in neutral rotation was checked in a qualitative fashion. Again, the stability of the glenohumeral joint depends on osseous concavity and ligamentous tension. We acknowledge that testing one mechanism in a quantitative way but the other in a qualitative way is a drawback of this study.

*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding source was the National Institutes of Health.

Investigation performed at the Biomechanics Laboratory and the Department of Orthopedics, Mayo Clinic and Mayo Foundation, Rochester

1. Apreleva, M.; Hasselman, C. T.; Debski, R. E.; Fu, F. H.; Woo, S. L-Y.; and Warner, J. J. P.: A dynamic analysis of glenohumeral motion after simulated capsulolabral injury. A cadaver model. J. Bone and Joint Surg., 80-A: 474-480, April 1998.
2. Aston, J. W., Jr., and Gregory, C. F.: Dislocation of the shoulder with significant fracture of the glenoid. J. Bone and Joint Surg., 55-A: 1531-1533, Oct. 1973.
3. Black, K. P.; Lim, T.-H.; McGrady, L. M.; and Raasch, W.: In vitro evaluation of shoulder external rotation after a Bankart reconstruction. Am. J. Sports Med., 25: 449-453, 1997.
4. Gill, T. J.; Micheli, L. J.; Gebhard, F.; and Binder, C.: Bankart repair for anterior instability of the shoulder. Long-term outcome. J. Bone and Joint Surg., 79-A: 850-857, June 1997.
5. Habermeyer, P.; Gleyze, P.; and Rickert, M.: Evolution of lesions of the labrum-ligament complex in posttraumatic anterior shoulder instability: a prospective study. J. Shoulder and Elbow Surg., 8: 66-74, 1999.
6. Hovelius, L.; Eriksson, K.; Fredin, H.; Hagberg, G.; Hussenius, Å.; Lind, B.; Thorling, J.; and Weckström, J.: Recurrences after initial dislocation of the shoulder. Results of a prospective study of treatment. J. Bone and Joint Surg., 65-A: 343-349, March 1983.
7. Jobe, C. M.: Posterior superior glenoid impingement: expanded spectrum. Arthroscopy, 11: 530-536, 1995.
8. Kummel, B. M.: Fractures of the glenoid causing chronic dislocation of the shoulder. Clin. Orthop., 69: 189-191, 1970.
9. Laumann, U.: Kinesiology of the shoulder joint. In Shoulder Replacement, pp. 23-31. Edited by R. Kolbel, B. Helbig, and W. Blauth. Translated by T. C. Telger. Berlin, Springer, 1987.
10. Lazarus, M. D.; Sidles, J. A.; Harryman, D. T., II; and Matsen, F. A., III: Effect of a chondral-labral defect on glenoid concavity and glenohumeral stability. A cadaveric model. J. Bone and Joint Surg., 78-A: 94-102, Jan. 1996.
11. Lippitt, S. B.; Vanderhooft, J. E.; Harris, S. L.; Sidles, J. A.; Harryman, D. T., II; and Matsen, F. A., III: Glenohumeral stability from concavity-compression: a quantitative analysis. J. Shoulder and Elbow Surg., 2: 27-35, 1993.
12. Liu, S. H., and Boynton, E.: Posterior superior impingement of the rotator cuff on the glenoid rim as a cause of shoulder pain in the overhead athlete. Arthroscopy, 9: 697-699, 1993.
13. Lusardi, D. A.; Wirth, M. A.; Wurtz, D.; and Rockwood, C. A., Jr.: Loss of external rotation following anterior capsulorrhaphy of the shoulder. J. Bone and Joint Surg., 75-A: 1185-1192, Aug. 1993.
14. Malicky, D. M.; Soslowsky, L. J.; Blasier, R. B.; and Shyr, Y.: Anterior glenohumeral stabilization factors: progressive effects in a biomechanical model. J. Orthop. Res., 14: 282-288, 1996.
15. Matsen, F. A., III, and Kirby, R. M.: Office evaluation and management of shoulder pain. Orthop. Clin. North America, 13: 453-475, 1982.
16. Matsen, F. A., III, and Thomas, S. C.: Glenohumeral instability. In Surgery of the Musculoskeletal System, edited by C. McC. Evarts. Ed. 2, pp. 1439-1469. New York, Churchill Livingstone, 1990.
17. Matsen, F. A., III; Thomas, S. C.; Rockwood, C. A., Jr.; and Wirth, M. A.: Glenohumeral instability. In The Shoulder, edited by C. A. Rockwood, Jr., and F. A. Matsen, III. Ed. 2, pp. 611-754. Philadelphia, W. B. Saunders, 1998.
18. Montgomery, W. H., III, and Jobe, F. W.: Functional outcomes in athletes after modified anterior capsulolabral reconstruction. Am. J. Sports Med., 22: 352-358, 1994.
19. Morrey, B. F.; Itoi, E.; and An, K.-N.: Biomechanics of the shoulder. In The Shoulder, edited by C. A. Rockwood, Jr., and F. A. Matsen, III. Ed. 2, pp. 233-276. Philadelphia, W. B. Saunders, 1998.
20. Neer, C. S., II: Shoulder Reconstruction, pp. 273-362. Philadelphia, W. B. Saunders, 1990.
21. Norlin, R.: Use of Mitek anchoring for Bankart repair: a comparative, randomized, prospective study with traditional bone sutures. J. Shoulder and Elbow Surg., 3: 381-385, 1994.
22. Novotny, J. E.; Nichols, C. E.; and Beynnon, B. D.: Kinematics of the glenohumeral joint with Bankart lesion and repair. J. Orthop. Res., 16: 116-121, 1998.
23. Palmer, I., and Widén, A.: The bone block method for recurrent dislocation of the shoulder joint. J. Bone and Joint Surg., 30-B(1): 53-58, 1948.
24. Paulos, L. E.; Evans, I. K.; and Pinkowski, J. L.: Anterior and anterior-inferior shoulder instability: treatment by glenoid labrum reconstruction and a modified capsular shift procedure. J. Shoulder and Elbow Surg., 2: 305-313, 1993.
25. Poppen, N. K., and Walker, P. S.: Normal and abnormal motion of the shoulder. J. Bone and Joint Surg., 58-A: 195-201, March 1976.
26. Protzman, R. R.: Anterior instability of the shoulder. J. Bone and Joint Surg., 62-A: 909-918, Sept. 1980.
27. Rowe, C. R.; Patel, D.; and Southmayd, W. W.: The Bankart procedure. A long-term end-result study. J. Bone and Joint Surg., 60-A: 1-16, Jan. 1978.
28. Speer, K. P.; Deng, X.; Borrero, S.; Torzilli, P. A.; Altchek, D. A.; and Warren, R. F.: Biomechanical evaluation of a simulated Bankart lesion. J. Bone and Joint Surg., 76-A: 1819-1826, Dec. 1994.
29. Symeonides, P. P.: The significance of the subscapularis muscle in the pathogenesis of recurrent anterior dislocation of the shoulder. J. Bone and Joint Surg., 54-B(3): 476-483, 1972.
30. Thomas, S. C., and Matsen, F. A., III: An approach to the repair of avulsion of the glenohumeral ligaments in the management of traumatic anterior glenohumeral instability. J. Bone and Joint Surg., 71-A: 506-513, April 1989.
31. Tijmes, J.; Loyd, H. M.; and Tullos, H. S.: Arthrography in acute shoulder dislocations. Southern Med. J., 72: 564-567, 1979.
32. Ungersböck, A.; Michel, M.; and Hertel, R.: Factors influencing the results of a modified Bankart procedure. J. Shoulder and Elbow Surg., 4: 365-369, 1995.
33. Urayama, M.; Itoi, E.; Konno, N.; Kido, T.; Sano, A.; and Sashi, R.: Quantitative assessment of capsular elongation in recurrent anterior dislocation of the shoulder using MR arthrography [abstract]. J. Shoulder and Elbow Surg., 7: 328, 1998.
34. Walch, G.; Boileau, P.; Noel, E.; and Donell, S. T.: Impingement of the deep surface of the supraspinatus tendon on the posterosuperior glenoid rim: an arthroscopic study. J. Shoulder and Elbow Surg., 1: 238-245, 1992.
35. Wirth, M. A.; Blatter, G.; and Rockwood, C. A., Jr.: The capsular imbrication procedure for recurrent anterior instability of the shoulder. J. Bone and Joint Surg., 78-A: 246-259, Feb. 1996.
Copyright 2000 by The Journal of Bone and Joint Surgery, Incorporated