Rotator cuff tear is the most common disease in the shoulder joint, with as many as 17 million persons in the United States at risk for disability1. The prevalence of rotator cuff disorders increases with advanced age. In addition, specific age groups are at high risk for the presence of full-thickness rotator cuff tear, and these patients should be monitored for tear progression1. Some researchers have proposed that many rotator cuff tears start at the anterior insertion of the supraspinatus muscle and extend posteriorly2-8. However, Kim et al.9 recently reported that degenerative rotator cuff tears initiate in a region 13 to 17 mm posterior to the biceps tendon and extend anteriorly and posteriorly from there. They suggested that this area corresponds to the center of the rotator crescent or rotator cable and also corresponds to the area of poorest vascularity of the rotator cuff tendon. Nonoperative treatment is the first line of management of rotator cuff disorders and is still effective for some patients7. Early detection and intervention, including surgical treatment, has been shown to improve the outcome of patients with this disease10-13. However, from a biomechanical perspective, the critical stage of rotator cuff tear at which surgical intervention is recommended is unknown. Most previous studies that evaluated the effect of rotator cuff tear size have focused on biomechanical changes with shoulder abduction14-18. Even though these studies are important for understanding abduction capability, other functional positions that are common for activities of daily living should be evaluated. Furthermore, most reported tests were performed by loading the rotator cuff and deltoid muscle, but these tests did not include the parascapular muscles, such as the pectoralis major or latissimus dorsi, which are very important for coordinated shoulder motion.
The purpose of this study was to determine if a critical rotator cuff tear stage exists that alters glenohumeral joint biomechanics throughout the rotational range of shoulder motion, and to evaluate the effect of parascapular muscle-loading on glenohumeral joint biomechanics with rotator cuff tear progression. Our hypotheses were that a critical rotator cuff tear stage will exist that leads to abnormal glenohumeral joint biomechanics, specifically rotational range of motion, path of the humeral head apex, and abduction capability, and that the pectoralis major and latissimus dorsi muscles will act as a stabilizer of the humeral head in a shoulder with a large or massive tear.
Specimen Preparation
Eight fresh-frozen human cadaver shoulders with a mean age of fifty-four years (range, thirty-four to sixty-nine years) were used. There were six male and two female specimens, and four right and four left shoulders. Specimens with a rotator cuff tear, glenohumeral joint contracture, or fracture of the acromion were excluded. The specimens were stored at –20°C until the day before testing, and each specimen was thawed overnight at room temperature in preparation for dissection and testing. The specimens were kept moistened with physiologic saline solution to prevent dehydration during testing. All soft tissues were removed except the glenohumeral joint capsule, coracoacromial ligament, and coracohumeral ligaments. The supraspinatus, infraspinatus, teres minor, subscapularis, deltoid, pectoralis major, and latissimus dorsi muscles were released from their origins, but for each of these muscles, the original insertion on the humerus was retained. The glenohumeral joint was vented by a small incision through the rotator interval in order to eliminate variability between specimens with regard to intra-articular pressure.
Suture loops were made with use of a modified Kessler stitch at the insertion of each muscle with number-2 FiberWire (Arthrex, Naples, Florida). In order to load the muscles anatomically, each muscle was divided into subregions on the basis of muscle fiber orientation, for seventeen lines of action: anterior and posterior regions for the supraspinatus; superior, middle, and inferior regions for the subscapularis, infraspinatus and teres minor, pectoralis major, and latissimus dorsi muscles; and anterior, middle, and posterior regions for the deltoid muscle.
Three reference screws were inserted on the scapula (at the coracoid, the anterior aspect of the acromion, and the posterior aspect of the acromion) and the humerus (at the proximal aspect of the bicipital groove, the distal aspect of the bicipital groove, and the greater tuberosity) to provide consistent digitization markers to define local coordinate systems on each bone for kinematic measurement. An aluminum rod was inserted into the medullary canal of the humeral shaft and secured with several crossing, interlocking screws. The scapula was fixed in the anatomical resting position with 20° of anterior tilt in the sagittal plane19,20. The aluminum rod was placed in a custom device that allowed axial rotation of the humerus and shoulder abduction (Figs. 1-A and1-B). Humeral axial rotation was defined on the basis of the anatomical relationship between the bicipital groove and the anterolateral corner of the acromion, as determined from a previous study21. When the bicipital groove was aligned with the anterolateral corner of the acromion at 90° of shoulder abduction, the humeral rotation was defined as 20° of external rotation.
Once the scapula and humerus were secured, origin plates for the rotator cuff and deltoid muscle were aligned to the scapular body. The origin plates for the pectoralis major and latissimus dorsi muscles were adjusted so that they were aligned anatomically relative to the clavicle and the inferior scapular border. For all seventeen lines of action, a fishing line was tied to each tendon suture and directed through the hole in the origin plate for anatomical alignment based on muscle fiber orientation. The amount of muscle-loading was determined on the basis of the physiological muscle cross-sectional area ratios22,23: 20 N for the supraspinatus, 30 N for the subscapularis, 30 N for the infraspinatus and teres minor, 60 N for the deltoid, 30 N for the pectoralis major, and 30 N for the latissimus dorsi. In order to evaluate the effects of parascapular muscles, three muscle-loading conditions were performed: rotator cuff only; rotator cuff with deltoid muscle; and rotator cuff, deltoid, pectoralis major, and latissimus dorsi muscles.
Rotator Cuff Tear Stages
Although a recent paper reported that rotator cuff tears start 13 to 17 mm posterior to the biceps tendon and extend anteriorly and posteriorly from this point9, the tear model used in this study follows the traditional concept that rotator cuff tears initiate anteriorly and extend posteriorly; therefore, a progressive rotator cuff tear model by sequential sectioning from the anterior rotator cuff based on footprint insertion anatomy was used (Fig. 2). Stage I was a tear of the anterior cord of the supraspinatus including the anterior rotator cable insertion24. Stage II was a tear of the entire supraspinatus. Stage III was a tear of the entire supraspinatus with detachment of 50% of the infraspinatus, and Stage IV was a tear of the entire supraspinatus and infraspinatus. The intact rotator cuff was used as a control. During the experiment, the tear sizes were monitored throughout all testing and the tear size did not increase due to testing procedures.
Biomechanical Testing
Testing was performed in the scapular plane (30° anterior to the coronal plane) at 0°, 30°, and 60° of shoulder abduction and a 2:1 ratio of glenohumeral to scapulothoracic abduction. The specimen was positioned at 0° of shoulder abduction with rotator cuff muscle-loading, and the maximum internal and external rotations were measured with 3.4 Nm of torque with use of a torque wrench attached to the humeral intramedullary aluminum rod. Prior to testing at each step, all muscles were loaded and the specimen was preconditioned by taking it from the maximum internal rotational position through the maximum external rotational position for five cycles to minimize the effect of soft-tissue viscoelasticity. The local coordinates of both the glenoid and the humerus were digitized with use of a MicroScribe 3DLX (Revware, Raleigh, North Carolina; accuracy within <0.3 mm) from maximum internal to maximum external rotation positions in 30° increments. The same procedure was then repeated for each muscle-loading condition. Following testing at 0° of shoulder abduction, testing was repeated at 30° and 60° of shoulder abduction.
The abduction capability was determined as the abduction angle achieved by increasing the middle deltoid load with the pectoralis major and latissimus dorsi muscles unloaded. The initial position of the shoulder was set to 10° of abduction with 0° of humeral axial rotation, and the humerus was allowed to rotate during abduction. The middle deltoid force was increased from 30 N to 70 N in increments of 10 N, and the abduction angle achieved was measured with use of the MicroScribe 3DLX.
After testing the intact cuff condition, sequential tear stages were made from Stage I to Stage IV and the same testing procedures were repeated with each muscle-loading condition in each shoulder abduction angle. Following all testing procedures, the specimens were carefully disarticulated and the humeral head and glenoid geometry were digitized with use of the MicroScribe 3DLX to calculate the position of the humeral head apex25, which is the most prominent point of the humeral head with respect to the geometric center of the glenoid. By tracing the path of the humeral head apex on the reconstructed glenoid, we could evaluate the rotational kinematics of the shoulder joint. Also,the differences in the location of the humeral head apex from normal and each tear stage were calculated in the anterior-posterior, superior-inferior, and medial-lateral directions.
Data Analysis and Statistics
All measurements were performed twice, and the averages were used for data analysis. A repeated-measures analysis of variance with a Tukey post hoc test (Statistica; StatSoft, Tulsa, Oklahoma) was used to determine significant differences between tear stages and muscle-loading conditions. The level of significance was set at p < 0.05.
Source of Funding
The study was funded in part by Veterans Affairs Rehabilitation Research and Development and Medical Research. This funding organization did not play a role in the investigation.
Rotational Range of Motion
Maximum external rotation was significantly increased at each abduction angle for all muscle-loading conditions after a Stage-II tear (p < 0.05, Table I). Maximum internal rotation was significantly increased only after a Stage-IV tear at each abduction angle with rotator cuff and deltoid muscle-loading and with rotator cuff, deltoid, pectoralis major, and latissimus dorsi muscle-loading (p < 0.05).
Abduction Capability
With lower deltoid loading conditions (<50 N), the abduction angle achieved by middle deltoid loading decreased significantly after a Stage-I tear (p < 0.05 vs. intact); however, in the higher deltoid loading conditions (60 N and 70 N), the abduction angle decreased significantly after a Stage-II tear (p < 0.05 vs. intact and p < 0.05 vs. Stage I [70 N],(Fig. 3).
Humeral Head Apex Position Compared with Intact Cuff State
The humeral head apex shifted 0.5 to 3.7 mm posteriorly after a Stage-III tear at the midrange of rotation (30° of internal rotation, 0° [neutral rotation], and 30° of external rotation) in 0° and 30° of abduction with rotator cuff muscle-loading and rotator cuff and deltoid muscle-loading conditions (p < 0.05, Fig. 4). At maximum internal rotation, the humeral head apex shifted 0.4 to 3.2 mm superiorly after a Stage-III tear for all muscle-loading conditions in 30° and 60° of abduction (p < 0.05, Fig. 5). In 0° of abduction, there was a superior shift (0.7 to 2.0 mm) only after a Stage-IV tear (p < 0.05). Following a Stage-IV tear, the humeral head apex shifted 1.3 to 3.5 mm laterally at maximum internal rotation for all muscle-loading conditions at each abduction angle (p < 0.05).
Role of Pectoralis Major and Latissimus Dorsi Muscles
Maximum internal rotation significantly decreased with the addition of deltoid loading to rotator cuff loading (p < 0.05) in each abduction position and in all tear stages. There was a tendency to be further decreased with additional pectoralis major and latissimus dorsi loading in each abduction position and in all tear stages, but it was not significant (p > 0.05). Deltoid, pectoralis major, and latissimus dorsi muscle-loading decreased maximum external rotation at all abduction angles for Stage-IV tear and for each rotator cuff tear stage except Stage I at 60° of abduction (p < 0.05) (Table I).
Deltoid muscle-loading elevated the humeral head apex (0.1 to 6.6 mm) at every rotational range of motion in 0° and 30° of abduction, regardless of the tear stage. With increasing tear stage, pectoralis major and latissimus dorsi muscle-loading reversed this abnormal migration (Fig. 6); that is, it reversed posterior shift of the humeral head apex at the midrange of rotational range of motion in 0° of abduction (p < 0.05), it reversed superior shift of the humeral head apex at maximum internal rotation in 0° of abduction (p < 0.05), and it reversed lateral shift of the humeral head apex at maximum internal rotation in 60° of abduction (p < 0.05).
The current study is a novel biomechanical analysis of rotator cuff tear progression in which glenohumeral kinematics were evaluated throughout the rotational range of motion at varying abduction angles and with anatomically based muscle-loading 21 that included the pectoralis major and latissimus dorsi muscles. Dividing one muscle into multiple lines of action allowed us to not only anatomically but also biomechanically create most representative rotator cuff tears, simulating the loss of muscle-force transmission at each stage of tear on the basis of the footprint anatomy. Quantitative evaluation of the role of the shoulder muscles was performed with use of a combination of rotator cuff, deltoid, and pectoralis major and latissimus dorsi muscle-loading.
Early detection of a rotator cuff tear, followed by proper management, may prevent detrimental biomechanical alterations and improve patient outcome. Clinically, some patients with small rotator cuff tears have severe painful dysfunction in activities of daily living and the tears extend quickly to become larger tears, whereas some patients may remain asymptomatic even with massive cuff tears. The symptoms may also be transient, intermittent, or inconstant during the natural history of a rotator cuff tear. Zingg et al.7 retrospectively reviewed nineteen patients who received nonoperative treatment for a massive cuff tear. After a mean duration of follow-up of four years, the patients still maintained satisfactory function despite significant progression of degenerative joint arthritis and fatty infiltration of rotator cuff muscles, a condition that is known to be irreversible26 and the presence of which is another important prognostic factor for the success of rotator cuff repair11. Since tear size and symptoms are not necessarily clinically correlated, it is difficult for orthopaedic surgeons to determine the appropriate time to recommend surgical repair of rotator cuff tears.
The findings from our present study suggest that there is a critical tear size at which abnormal glenohumeral joint kinematics become apparent, and our findings also provide insight into the possible mechanisms associated with tear progression or subsequent pathology. The critical tear size for humeral rotational motion and glenohumeral abduction capability was a tear of the entire supraspinatus. Decreased abduction capability due to a tear of the entire supraspinatus may be explained by the function of the muscle as an initiator of shoulder abduction. This finding is also supported by the findings of Halder et al.14, who reported that small tears that do not violate the rotator cable presented no loss of shoulder strength, as well as by the findings of Parsons et al.17, who found that glenohumeral joint reaction forces were significantly decreased with rotator cuff tears that extended beyond the supraspinatus tendon.
Rotator cuff tear progression to half of the infraspinatus, described as a Stage-III tear, was the critical tear size for significant changes in rotational humeral head kinematics: specifically, superior and lateral shift of the humeral head at maximum internal rotation and posterior shift at the midrange of rotational motion. These results may be due to the pull from the unbalanced force-couple relationship between the subscapularis and infraspinatus muscles because of tear progression in the infraspinatus (Figs. 7-A and 7-B), loss of the muscle barrier effect, and decreased joint compressive force. This concept is supported by Keener et al.8, who reported that tears extending into the infraspinatus are associated with greater humeral migration than is seen with isolated supraspinatus tears. Rotator cuff tears that extend into the infraspinatus tendon may be too large to allow compensatory contraction through the rotator cuff cable, resulting in humeral head migration. Mura et al.27 also reported that the infraspinatus contributed to abduction torque generation, and stabilized the humeral head against elevation. The results of our study further demonstrate the importance of the infraspinatus tendon in maintaining normal glenohumeral joint kinematics, especially during humeral rotation.
With regard to the involvement of the rotator cable, there were no significant biomechanical changes in association with small rotator cuff tears that were limited to the anterior portion of the supraspinatus (a Stage-I tear) in the current tear model. Although the anterior origin of the rotator cable was disrupted in a Stage-I tear, the biomechanical alterations might be avoided if the remaining part of the rotator cable is intact and maintaining the force couple. However, with propagation of a tear to the entire supraspinatus (a Stage-II tear), the disruption of the rotator cable progressed and biomechanical alterations developed. Further tear extension to the infraspinatus (a Stage-III tear) led to further deterioration of the rotator cable construct and force couple and to disruption of glenohumeral kinematics. Small tears of the rotator cable had little effect on glenohumeral biomechanics, but progressive disruption of the rotator cable would result in serious biomechanical consequences, regardless of the initial location of the rotator cuff tear.
Previous biomechanical studies measured humeral head translation15, glenohumeral joint reaction force17, and glenohumeral abduction torque2-4 with simulated loading of the rotator cuff only or the rotator cuff plus the deltoid. Previous studies have also shown that rotator cuff tears lead to superior translation of the humeral head when the superior forces generated by the deltoid are no longer effectively opposed14-17,27,28. Loading the rotator cuff and deltoid only, without consideration of the pectoralis major and latissimus dorsi muscles, may not represent the physiological condition of the force couple in the superior-inferior direction. Due to their large cross-sectional areas and moment arms, the pectoralis major and latissimus dorsi muscles may significantly influence the kinematics of the shoulder joint. In this study, loading the pectoralis major and latissimus dorsi muscles restored abnormal migrations of the humeral head apex. This result suggests that the pectoralis major and latissimus dorsi muscles play a role as a humeral head depressor in 0° of abduction and a humeral head compressor in 60° of abduction. These findings are corroborated by Steenbrink et al.29, who reported that in patients with massive cuff tears, arm adductors including the pectoralis major and latissimus dorsi muscles would be coactivated at the cost of arm force and abduction torque in order to reduce the superior translation force. Our findings support the use of rehabilitation for strengthening the remaining intact musculature in patients whose painful rotator cuff tear is managed nonoperatively, possibly decreasing pain by improving glenohumeral kinematics.
There are several limitations to the current study. First, this was a cadaver study, so pain, which may have a significant effect on glenohumeral kinematics or rotational range of motion, cannot be considered. A very small tear, which would otherwise not have biomechanical consequences, may be associated with significant kinematic changes secondary to pain-related inhibition of muscle activity. Conversely, a larger rotator cuff tear, which should be associated with more biomechanical alterations, is sometimes tolerated well due to an unexplained absence of pain. Also, in contrast to the positions studied during in vivo tracking30-32 and ex vivo robotic arm studies33-35, we chose several clinical rotator-cuff-relevant shoulder positions that are common to activities of daily living and evaluated the effect of tear progression during humeral rotational range of motion. Because the arc of abduction in patients with massive rotator cuff tear has been shown to be significantly decreased, our tests included as much as 60° of abduction. Also, because these studies are quite complex and labor-intensive as well as extremely time-consuming, we limited the positions that were tested and, since our primary focus was the effect of humeral rotation, we limited our study to the scapular plane. Second, there is concern that each subsequent measurement may magnify the effect of the rotator cuff tear via tissue creep, as the measurements were performed in a nonrandomized fashion. This was avoided by preconditioning of the specimen, keeping the specimen moistened with normal saline solution, monitoring the extension of tear, and performing two trials of each data point to assess repeatability. During the experiment, the difference in the glenohumeral joint kinematics between two trials was <1 mm. Finally, constant muscle-loading for all positions and conditions was used. As there are an infinite number of different muscle force combinations for any given glenohumeral joint position, we chose muscle-loading conditions on the basis of the ratio of physiological cross-sectional areas to simulate muscle co-contraction.
Despite these limitations, the current study was a comprehensive biomechanical analysis of rotator cuff tear progression to determine the critical tear stage that significantly affects glenohumeral joint biomechanics throughout the rotational range of motion and glenohumeral abduction with anatomically based muscle-loading, including the rotator cuff, deltoid, pectoralis major, and latissimus dorsi muscles.
In conclusion, a tear of the entire supraspinatus was the critical stage for increasing shoulder external rotational motion and for decreased shoulder abduction capability, and tear progression to the infraspinatus was the critical stage for significant changes in humeral head kinematics. The pectoralis major and latissimus dorsi muscles played an important role to depress and stabilize humeral head migration due to deltoid loading. The findings from our study suggest that differential surgical intervention might be considered for patients in whom a rotator cuff tear extends beyond these biomechanical critical tear stages, with the primary goal of restoring normal glenohumeral joint biomechanics to prevent detrimental rotator cuff tear progression and subsequent pathology, based on our findings of differential loss of function according to tear progression. For patients whose rotator cuff tear is managed nonoperatively, our findings support the use of rehabilitation for strengthening the remaining intact musculature, especially the pectoralis major and latissimus dorsi muscles. The strengthening of these muscles may improve the functional outcome by restoring glenohumeral joint kinematics and possibly reduce pain. Early detection of rotator cuff tear, followed by proper management, may prevent detrimental biomechanical alterations and improve patient outcome.