Massive rotator cuff tears (>5 cm1) generally involve complete rupture of the supraspinatus tendon and partial or complete rupture of one or more of the subscapularis, infraspinatus, and teres minor tendons. In one report, eleven of 105 rotator cuff tears diagnosed in patients with symptomatic shoulders were categorized as massive2. An anatomic cadaver study demonstrated that two of fifty-seven rotator cuff tears were massive, involving the supraspinatus, infraspinatus, and subscapularis3.
Surgical repair of rotator cuff tears is a common procedure. Both open and arthroscopic surgical techniques to reapproximate the rotator cuff tendons to the humerus are performed at a considerable cost to society. However, recent studies have demonstrated high rates of recurrence after the repair of large and massive tears4,5. Despite recurrent tears, many patients in those series maintained excellent shoulder function. Additionally, some rotator cuff tears are asymptomatic6 and are only identified incidentally. Good results have been reported following débridement, decompression, and/or partial repair7-9.
These observations illustrate that our understanding of shoulder function in the presence of a substantial tear is not complete. One hypothesis is that contraction of the deltoid muscle causes the humeral head to be captured underneath the coracoacromial arch, with subsequent pivoting of the humerus about this abnormal center of rotation10. DePalma et al. stated that good function is compatible with massive avulsion of the rotator cuff, provided that balance between the deltoid muscle and the remaining intact portion of the rotator cuff is not severely impaired6. Burkhart et al., stressing the importance of the transverse force couple, stated that unrepaired rotator cuff tears with intact force couples and an intact rotator cable may be compatible with normal function if the patient can obtain satisfactory pain relief, although these tears may be quite large9. These statements seem to agree with the observation that the factors more likely to result in satisfactory postoperative range of motion and overhead function were intact subscapularis and teres minor tendons with absence of muscular atrophy8.
The kinematics of the normal glenohumeral joint approximates a spherical joint11 with a small amount of translation12,13. Abnormal superior head translation has been associated with rotator cuff tears in vivo10,13,14 and in cadavera15-17. The primary destabilizing force is the deltoid, causing superior translation of the humeral head. The rotator cuff provides dynamic stability to maintain spherical joint kinematics. The rotator cuff muscles have been described functionally as humeral head rotators, which position the upper extremity; humeral head depressors18, which impart stability through concavity-compression19; and humeral head steerers, which actively control the fulcrum of the glenohumeral joint during motion20. Rotator cuff electromyographic activity precedes deltoid and pectoralis major electromyographic activity for volitional movement21, suggesting that the rotator cuff muscles dynamically stiffen the glenohumeral joint in preparation for the larger destabilizing deltoid and pectoralis major contractions.
Reflex arcs have been identified in the feline shoulder22. In addition, histologic studies and somatosensory cortical evoked potentials have provided evidence of mechanoreceptors and proprioceptive pathways involving capsuloligamentous and tendinous structures of the human shoulder23. Disturbance of these pathways has been implicated in pathologic shoulder conditions such as instability24 and functional impingement resulting from neuromuscular insufficiency of the rotator cuff25. In the case of instability, there have been reports of normalization of proprioception after surgical stabilization24.
We hypothesized that, in the presence of a massive rotator cuff tear, the necessary force levels could be developed in the remaining intact portions of the teres minor, inferior infraspinatus, and inferior subscapularis to enable glenohumeral abduction while maintaining superior translation of the humeral head within normal limits. Additionally, we hypothesized that a greater role by the rotator cuff muscles, i.e., an increased rotator cuff force, would obviate the need for increased deltoid force.
Specimen Preparation
Six fresh-frozen human cadaver upper extremities from five male and one female donor who had had a mean age of 69.2 years (range, forty-one to 101 years) at the time of death were selected for the study. Two specimens were from the left side, and four were from the right side. All six specimens consisted of the proximal part of the humerus and the scapula. The mean anterior-posterior width of the specimen as measured from the lesser tuberosity to the middle of the insertion of the infraspinatus was 49.1 mm (range, 43.8 mm to 53.6 mm). All six specimens were free of evidence of previous surgery and preexisting pathologic conditions.
Superficial soft tissues were resected, with the rotator cuff and the coracohumeral ligament being spared. The tendons of the rotator cuff were transected 2 cm proximal to their myotendinous junctions. The insertion of the deltoid was identified and marked prior to resection. The humerus was amputated just distal to the deltoid tuberosity. The glenohumeral joint was vented to atmosphere during dissection, and normal saline solution was periodically applied during testing to prevent desiccation. A rongeur was used to perform an acromioplasty on each specimen, with resection of one-half of the thickness of the acromion and the coracoacromial ligament. This provided sufficient clearance such that contact between the rotator cuff and the acromion did not occur during testing, allowing investigation of the isolated glenohumeral articulation without confounding articulation with the acromion.
Shoulder Controller
A custom shoulder controller was designed and built to mimic position control characteristics of in vivo neuromuscular control. This apparatus used eight independent stepper motors connected with Spectra cord (braided ultra-high molecular weight polyethylene fiber selected for its high strength, low stretch, and low friction) to the rotator cuff tendons and deltoid insertion, replicating the actions of the deltoid, supraspinatus, infraspinatus, teres minor, superior subscapularis, and inferior subscapularis muscles. For the rotator cuff tendons, the cord was sutured directly with use of the Krackow technique. The cords were directed through a series of pulleys and bushings to approximate the origins of the muscles that they emulated (Fig. 1). The pulleys and bushings were positioned at the center of the muscle origins that were identified for each specimen during dissection26-28. The bushings were lined with fluorinated ethylene-propylene. The tension in each cord that was generated by the stepper motors was measured with use of force transducers (Transducer Techniques, Temecula, California).
The scapula was rigidly mounted with the medial border fixed perpendicular to the floor and inclined 10° anteriorly in the sagittal plane, approximating the anatomic shoulder position at rest. All structural components of the apparatus were nonmetallic, eliminating potential interference with the electromagnetic three-dimensional tracking system. Equivalent mass distribution of the upper extremity with the elbow fully extended and the wrist in neutral was simulated by attaching a 3.5-kg mass made of Garolite to a fiberglass rod secured in the distal part of the humerus such that the mass was 318 mm distal to the greater tuberosity29-31 (Fig. 1).
Humeral Head Position, Center of Rotation, and Three-Dimensional Tracking
A FASTRAK device (Polhemus, Colchester, Vermont) was used to track the position of the humeral head and the orientation of the humerus in three-dimensional space. The manufacturer's stated accuracy of this device is 0.0381 mm and 0.15° in metal-free environments. The FASTRAK transmitter and receiver were mounted to the scapular mount and the humerus, respectively.
The center of rotation for each specimen was determined in a nondestructive manner at the beginning of each test by acquiring FASTRAK data during circumduction. For this maneuver, the humeral head was held in a reduced position with a manually applied compressive force while the humerus was passively ranged through approximately 90° in the scapular plane, 45° in the flexion-extension plane, and 45° of axial rotation. A least-squares sphere-fitting routine was used to identify the center of rotation32,33. This point was defined as the functional center of rotation, and all rotations and translations were referenced to this point throughout the experiment. As a check of this dynamic method, the articular surface of the disarticulated humeral head was digitized after testing with use of the FASTRAK stylus and the geometric center of the best-fit sphere was calculated. The mean difference between the functional center and the geometric center was 2.2 mm (range, 1.2 to 3.3 mm), confirming that this dynamic technique was appropriate.
Registration of the three-dimensional positions of the muscle origins and insertions with use of the FASTRAK digitizer allowed tracking of the individual muscle force vectors throughout the experiment. With use of the conditions of static equilibrium, the joint reaction force was calculated from these force vectors and the gravitational force.
Data Acquisition and Control
A personal computer running a LabVIEW virtual instrument (National Instruments, Austin, Texas) was used for data acquisition from the force transducers and the FASTRAK. The computer also contained a hardware interface to control the stepper motors. The custom software employed independent closed-loop feedback to control each stepper motor. The control strategy that we utilized was chosen because it replicated abduction in the plane of the scapula and the constraints and muscle forces that were generated were within physiologic limits.
The desired abduction angle was entered into the computer, the FASTRAK monitored the actual abduction angle, and the feedback control loop for the middle deltoid stepper motor automatically adjusted the middle deltoid force until the humerus achieved the desired glenohumeral abduction angle. The posterior deltoid was loaded with the same instantaneous force as the middle deltoid, and the anterior deltoid force was adjusted to maintain abduction in the plane of the scapula. The biceps tendon was not loaded in either the intact or rotator cuff tear conditions in this study.
The controller limited superior translation to 3.0 mm and maintained neutral axial rotation as follows. As the controller increased the force in the deltoid to move toward the desired abduction angle, the shear force of the deltoid caused the humeral head to translate slightly superiorly. This superior translation was detected by the FASTRAK system, and the controller increased the subscapularis force (shared equally between the superior and inferior divisions) and the supraspinatus force (for the intact condition) to counteract this motion. The increasing subscapularis force caused the humerus to begin to rotate internally. This rotation was detected by the FASTRAK system, and the controller increased infraspinatus and teres minor forces (shared equally) to restore neutral rotation. There were eight independent closed-loop feedback control loops running simultaneously at thirty iterations per second to achieve the desired angle and translation positions. Thus, the force magnitudes were not specified a priori; rather, they were determined empirically by the control loops.
Each specimen was tested in the intact condition followed by three simulated progressive rotator cuff tear conditions. The simulated tears were produced by resecting a crescent-shaped portion of rotator cuff tendon adjacent to the medial extent of the cuff insertion. The medial-lateral resection width was 1.5 cm. In the anteroposterior direction, the simulated tear involved the entire supraspinatus and rotator interval with equal extension into the subscapularis anteriorly and the infraspinatus posteriorly for total widths of 6, 7, and 8 cm. Standardized absolute tear sizes were chosen to correlate with common clinical terminology that describes rotator cuff tears in terms of width, independent of humeral head size. To simulate a chronic tear, the supraspinatus was not tensioned for the rotator cuff tear conditions. Static data were recorded at 5° intervals from 10° to 80° of glenohumeral abduction. Five trials were performed for each of the four rotator cuff tear conditions.
It is generally accepted that the effects of fatigue become a factor as more fast-twitch fibers are recruited when >50% of maximum force is utilized34. To assess whether the increased force requirements associated with the simulated tears placed the muscles at risk for fatigue, the force requirements were compared with the estimated fatigue threshold.
Statistical Methods
A repeated-measures four-factor analysis of variance with random effects was performed to test the effects of cuff tear, abduction angle, specimen, and trial on subscapularis force, combined infraspinatus and teres minor force, total deltoid force, and joint reaction force. Bonferroni post hoc analysis was performed to test for significant differences between cuff condition groups. Because four variables were evaluated for significance, the level of alpha was decreased to 0.01, decreasing the probability of a type-I error due to multiple comparisons. The MATLAB statistical analysis toolkit (The MathWorks, Natick, Massachusetts) was used to perform all statistical analyses.
Stable glenohumeral abduction from 10° to 80° in the scapular plane was achieved for all six specimens while limiting superior translation of the humeral head to =3.0 mm. For the intact condition, the subscapularis and combined infraspinatus/teres minor forces both peaked at 45° of abduction with a mean maximum force (and standard deviation) of 55 ± 12 N for the subscapularis (Fig. 2) and 62 ± 12 N for the combined infraspinatus and teres minor (Fig. 3). The effect of the simulated massive rotator cuff tears was to increase the maximum rotator cuff force required for abduction (p < 0.01). Furthermore, the maximum force occurred earlier in the abduction arc as tear sizes increased. Compared with the forces required for the intact rotator cuff, the subscapularis required 30% (16.7 N/55.4 N), 44% (24.6 N/55.4 N), and 85% (47.4 N/55.4 N) force increases for the 6, 7, and 8 cm tears, respectively. The combined infraspinatus and teres minor required 32% (19.9 N/62.1 N), 45% (27.7 N/62.1 N), and 86% (53.7 N/62.1 N) force increases, respectively. Combined forces for the deltoid, superior and inferior subscapularis, and infraspinatus and teres minor were reported as scalar sums of the magnitudes of the respective forces rather than vector sums to reflect the total effort of the muscle rather than the resultant force vector generated by the muscle.
The increased force requirements are illustrated relative to the fatigue threshold in Figure 4. The dashed line at the 100% increase level reflects the twofold increase in demand when going from 25% to 50% of maximum force; that is, the region above 100% indicates the conditions for which fatigue is expected to limit shoulder abduction. The graph illustrates that, up to 30° of abduction, there will be insufficient force-generating capability for the inferior cuff muscles to compensate for the 8-cm tear during repetitive activities as they are able to do for the 6-cm tear and, for the most part, for the 7-cm tear.
For abduction in the intact condition, the total required deltoid force increased with increasing abduction to a maximum of 241 ± 10 N at 80°. Each of the simulated massive rotator cuff tears required the deltoid force to increase earlier in the abduction arc, but the maximum deltoid force for the simulated tear conditions never exceeded the deltoid force required at maximum abduction for the intact condition (Fig. 5). Thus, despite maintaining spherical joint kinematics, greater deltoid force was required in the earlier range of abduction.
The calculated joint reaction force increased throughout the range of abduction, with greater forces being generated at higher abduction angles (Fig. 6). The effect of tear size was to increase the joint reaction force for abduction angles between 10° and 45°. The maximum joint reaction force for the intact condition (at 80°) was not exceeded.
The largest percentage increases in rotator cuff force requirements and deltoid force requirements resulting from the simulated rotator cuff tears occurred in early abduction (Table I). The effect of the 6-cm tears was an approximately twofold increase in the demand on the rotator cuff muscles compared with the demand on the deltoid. For larger tears, the effect was an approximately 2.5-fold increase in demand on the rotator cuff compared with the deltoid.
The present study demonstrates that glenohumeral abduction without excessive superior translation is possible in the presence of a massive rotator cuff tear, provided that the remaining intact portion of the cuff can produce sufficient force. This finding is in agreement with those in similar reports6,10,11,26. However, even with increased rotator cuff forces, the deltoid force requirement increased in early abduction. For some of the simulated tears, the magnitude of the increase in the required rotator cuff force was within theoretical physiologic limits of the muscles35,36. However, patient-specific factors may prevent a given patient from achieving such increased force. Because an acromioplasty was performed and the controller prevented superior humeral head translation of >3.0 mm, this experiment did not investigate the kinematics of shoulder abduction that relies on articulation of the humeral head with the acromion. Such "captured fulcrum kinematics" might be expected to decrease the rotator cuff force required for abduction10.
The control strategy used for moving and positioning the shoulder in the present study utilized closed-loop glenohumeral orientation and translation feedback control. Compared with other approaches11,16,37, this strategy has the unique advantage of determining the dynamic forces necessary for glenohumeral stability. The failure of biomechanical studies to take into account the dynamic contributions of the rotator cuff and other muscles to glenohumeral stability has been cited as a limitation of cadaveric modeling of shoulder instability38.
Subscapularis force requirements for stable glenohumeral abduction were increased 30% to 85%, depending on tear size. Clinical studies have corroborated the essential role of the subscapularis. In an electromyographic study of patients with two-tendon rotator cuff tears39, a trend of increased subscapularis activity in asymptomatic patients compared with symptomatic patients was reported. Those results correlate well with the increased subscapularis force requirement observed in the present study, suggesting that asymptomatic subjects possess effective subscapularis function. Likewise, subscapularis transfers that are performed to treat rotator cuff deficiency may preclude the subscapularis from performing its essential role and prevent spherical kinematics.
The rotator cuff force requirements increased substantially for the tear conditions compared with the intact condition, ranging from a 30% increase for the 6-cm tear to an 86% increase for the 8-cm tear. When considering rehabilitation, one must assess whether such substantial force increases are possible for a given patient. Concomitant partial repair may sufficiently decrease the cuff force requirement such that rehabilitation is effective. However, elevated forces in the cuff tendon may contribute to cuff degeneration as recent evidence has suggested that the natural history of rotator cuff tears is progression in size40.
As a result of the rotator cuff tear, increased force is applied through a smaller cross-section of the remaining intact portion of the tendon. While no complete rotator cuff ruptures occurred during testing, the present study did not investigate whether the cuff can withstand the resulting elevated intratendinous stresses over time. These elevated stresses may lead to progression of the tear, perhaps justifying early repair. As a specific rotator cuff tear configuration was chosen for the study, the results likely are not generalizable to other tear configurations.
When fatigue was factored into the performance of the rotator cuff muscles, the ability to initiate abduction was compromised for the simulated 8-cm tear and, to a lesser degree, for the 7-cm tear. Thus, repetitive activities that include the early abduction range can be expected to result in superior translation of the humeral head as a result of fatigue. The effect of fatigue on glenohumeral kinematics was documented in vivo in a radiographic study of men41.
In fulfilling its role as the major abductor of the glenohumeral joint, the deltoid also generates substantial destabilizing shear forces42, especially at low abduction angles. Static and dynamic stabilizing mechanisms normally counteract these shear forces. Shoulders in which the stability mechanism is compromised may not be able to abduct or may exhibit abnormal kinematic patterns such as "captured fulcrum kinematics."10 The present study demonstrated that in addition to restoring spherical kinematics by increasing rotator cuff forces, increased deltoid forces also were required at low-to-intermediate abduction angles (Table I). Relatively greater increases were observed for the rotator cuff compared with the deltoid. While rehabilitation of the deltoid is important, these results highlight the relatively greater importance of rehabilitation of the rotator cuff muscles to prevent superior humeral head translation and subacromial impingement. Indeed, emphasis on rotator cuff strengthening has become a mainstay of nonoperative shoulder treatment43.
The joint reaction force was calculated with use of the digitized lines of action of the rotator cuff and deltoid and the center of mass of the humerus rather than being measured directly. Analytical models have determined the joint reaction force to be approximately one-half body weight29. This value is in close agreement with the results of the present study. The elevated joint reaction forces in early abduction that result from rotator cuff tears may contribute to degeneration of the glenohumeral joint despite the restoration of spherical kinematics. This may provide additional justification for early repair of rotator cuff tears.
The anterior and posterior rotator cuff muscles exhibited 19% and 17% increases, respectively, in the force requirements in the early abduction range when the simulated tear was increased from 6 to 7 cm. The cuff muscles exhibited 51% increases for the anterior and posterior cuff force requirements after the simulated tear was increased from 7 to 8 cm. For the 8-cm-tear condition, the force requirements for the anterior cuff were up to three times the force requirements for the intact condition. It may not be realistic to expect these muscles to compensate to such a degree.
In summary, we have reported the results associated with a custom cadaver shoulder controller that utilized position and orientation closed-loop feedback control. Rotator cuff tear size, rotator cuff muscle force, and deltoid force are important determinants of shoulder function in the presence of massive rotator cuff tears. The findings of the present study provide guidance for the rehabilitation of patients with rotator cuff tears and those who have had rotator cuff repair and may provide justification for early surgical repair of rotator cuff tears. 