Twenty rotator cuff reruptures were identified with postoperative magnetic resonance imaging in a series of sixty-five consecutive patients who had had a repair of a full-thickness tear of the rotator cuff. Of the twenty patients who had a rerupture, thirteen were men and seven were women; the mean age at the time of the index operation was fifty-nine years (range, forty-seven to seventy-one years). The reruptures were diagnosed with the use of established magnetic resonance imaging criteria15,24,27. Initially, eight patients had had a tear of one tendon (the supraspinatus), ten had had a tear of two tendons (the supraspinatus and the subscapularis in five and the supraspinatus and the infraspinatus in five), and two patients had had a tear of three tendons. Seventeen patients had involvement of the dominant side (Table I).
Findings at the Index Operation
In eleven of the twenty patients, the quality of the tendon of the supraspinatus muscle had been judged to be fair or poor on the basis of unusually poor holding power for sutures or brittle, myxomatous macrostructure. After mobilization of the torn musculotendinous units, the tendons were grasped with nonresorbable sutures according to the modified Mason-Allen technique11,12. Both strands of one stitch were brought through the tuberosity and knotted over two holes of a thin titanium plate with seven rounded holes (Stratec Synthes SA, Oberdorf, Switzerland), which served as a cortical bone augmentation device11,12. The titanium plate does not cut the sutures and does not interfere with assessment of the operatively treated shoulder with magnetic resonance imaging. In one patient (Case 14), the supraspinatus tendon was unexpectedly found to be too far retracted and too degenerated to allow complete reinsertion on the greater tuberosity, so only a partial repair could be performed. An anterior acromioplasty was performed in sixteen of the twenty patients. An acromioplasty was avoided if postoperative anterosuperior subluxation seemed likely due to difficulties in obtaining an optimal closure of the rotator interval; this was the case for massive tears involving the supraspinatus and the subscapularis (four patients). Furthermore, five patients had a tenotomy or tenodesis of the tendon of the long head of the biceps muscle because it was either degenerated or partially ruptured. The decision to perform a tenotomy rather than a tenodesis was based on cosmetic considerations for the individual patient (there may be bulging of the biceps muscle after tenotomy). In four patients, the long head of the biceps tendon was already torn at the index operation.
Postoperative Rehabilitation
Ten patients wore an abduction splint on the arm for six weeks postoperatively because the intraoperative quality of the supraspinatus had been thought to be poor (five patients), the reinserted supraspinatus tendon had been under tension (three patients), or the index operation was a reoperation for a failed repair of a rotator cuff tendon (two patients). The remaining ten patients, most of whom had had a large subscapularis lesion, wore only a sling for six weeks. Immediately postoperatively, passive range-of-motion exercises were begun within the range that had been found to be safe intraoperatively. Active range-of-motion exercises were started after six weeks. Strengthening of the rotator cuff muscles was begun after twelve weeks.
Complications
No patient had hematoma, wound dehiscence, infection, stiffness necessitating manipulation, deltoid dehiscence, or neural injury. At the time of the most recent follow-up, only one patient (Case 8) had had a reoperation; this patient had had an arthroscopic débridement because of persistent pain.
Clinical Assessment
At the most recent follow-up evaluation, clinical assessment was performed, in a standardized fashion, for the purpose of this study by a single examiner (B. J.) who was not the operating surgeon (C. G.). The clinical assessment consisted of a structured interview and a detailed physical examination including all elements needed for assessment of shoulder function according to the system of Constant and Murley5,6. This score is based on a scale of 100 points. A maximum of 35 points is assigned for subjective variables (pain, activities of daily living, and functional use of the arm); 40 points, for range of motion; and 25 points, for quantitative measurement of abduction strength.
Active shoulder motion was measured with the patient sitting. The range of flexion was assessed in the sagittal plane as the angle between the humeral shaft and the midthoracic line (not the vertical). Abduction was always measured with simultaneous maximal abduction of both arms as the angle of the humeral shaft with the midthoracic line. Active external rotation was assessed, according to the method of Constant and Murley5,6, while the patient performed five functional external rotation movements without touching the head with the hand. Internal rotation was determined by the spinous process that the patient could reach with the thumb. Abduction strength was measured with the patient standing and the arm abducted to 90 degrees in the scapular plane with the elbow extended and the forearm pronated. The resistance of an Isobex dynamometer (Cursor SA, Bern, Switzerland) was applied to the wrist, and three consecutive measurements of a duration of five seconds each (the B mode of the device) were averaged to measure the strength. One point was attributed per 0.48 kilogram of strength measured6. If 90 degrees of abduction was not achieved, abduction strength was automatically considered to be zero. The total score obtained was also related to the age and gender-matched normal values that had been identified by Constant5, and the respective value was called the relative Constant score. In addition, the patients estimated the value of the operatively treated shoulder as a percentage of the value for an entirely normal shoulder (that is, 100 percent). This value was called the subjective shoulder value.
Although clinical data on both shoulders were collected for comparison of the treated and contralateral limbs, the high prevalence of symptomatic disease of the contralateral shoulder precluded the use of the contralateral side for meaningful comparison.
Radiographic Assessment
For all twenty patients, postoperative radiographic assessment included standardized radiographs (anteroposterior radiographs with the arm in neutral rotation and axillary lateral radiographs) and magnetic resonance imaging. Magnetic resonance imaging was performed with a 1.0-tesla scanner (Siemens, Erlangen, Germany). Continuity or rerupture of the tendon was assessed on coronal oblique T2-weighted and proton-density-weighted images as well as short inversion recovery sequences according to established magnetic resonance imaging criteria15,24,27. When a fluid-equivalent signal or nonvisualization of the supraspinatus, infraspinatus, or subscapularis tendon was found on at least one T2-weighted or fat-suppressed section, the diagnosis of a full-thickness rerupture was made27. Additionally acquired parasagittal T1-weighted turbo spin-echo magnetic resonance images parallel to the glenohumeral joint were obtained for qualitative and quantitative assessment of the rotator cuff muscles36. The slices covered the rotator cuff from the humeral tuberosities to the medial third of the scapula. Cross-sectional areas of the supraspinatus were measured on the most lateral image on which the scapular spine was in contact with the scapular body (Y-shaped view)36. Intramuscular fatty degeneration and atrophy of the muscle bellies were assessed as described by Goutallier et al.14 for computed tomography scanning and validated by Fuchs et al.9 for magnetic resonance imaging.
The preoperative radiographic assessment included magnetic resonance imaging (nineteen patients) or computed tomography scanning (one patient) and standard radiographs. In order to compare the preoperative and postoperative extents of the rotator cuff defect, the sizes of the tears were assessed on the basis of the maximal mediolateral and anteroposterior diameters. Preoperative and postoperative glenohumeral osteoarthritic changes were assessed on standard radiographs according to the classification of Samilson and Prieto31.
Statistical Methods
The Mann-Whitney U test was used for unpaired groups, and the Wilcoxon test was used for paired groups. The level of significance was set at p < 0.05. Spearman's correlation coefficient was used to test relationships between variables.
Imaging
The mean acromiohumeral distance measured on a true anteroposterior radiograph with the forearm in neutral rotation decreased significantly (p = 0.01), from ten millimeters (range, seven to thirteen millimeters) before the operation to 8.4 millimeters (range, four to twelve millimeters) at the time of follow-up at a mean of thirty-eight months (range, twenty-seven to forty-nine months) postoperatively. The distance was less than seven millimeters in three patients (one of the ten who initially had had a two-tendon tear and both of the patients who initially had had a three-tendon tear), reflecting a large rotator cuff tear34. On the axillary lateral radiograph, the humeral head appeared centered in nineteen of the twenty patients. In one patient (Case 17), who had sustained a traumatic posterior dislocation of the shoulder with the rotator cuff tear, the head was slightly subluxated posteriorly.
Osteoarthritic changes, as assessed according to the classification of Samilson and Prieto31, had increased significantly (p = 0.002) from the preoperative to the postoperative evaluation, but never by more than one stage. At the time of follow-up, stage-II osteoarthritis was found in four patients (three who had originally had a two-tendon tear and one who had had a three-tendon tear). Three of these four patients had a retear that was larger than the original tear. Overall, sixteen reruptures were smaller than the original tear and four (one in the one-tendon-tear group and three in the two-tendon-tear group) were larger than the initial tear. The mean decrease in the tear size from the preoperative to the postoperative evaluation was significant for the entire series of twenty patients (p = 0.045) and for the ten patients who originally had had a two-tendon tear (p = 0.032), but it was not found to be significant for the eight patients who had had a one-tendon tear (p = 0.078), probably because of the small number of patients.
Structural failure always involved the repaired tendon and always involved the supraspinatus (Fig. 1). It extended into an infraspinatus that had been intact at the time of the index operation in only two patients (Fig. 2).
Fatty muscle degeneration of the supraspinatus progressed in ten patients (two of the eight who had originally had a one-tendon tear, seven of the ten who had had a two-tendon tear, and one of the two who had had a three-tendon tear). Progression of fatty muscle degeneration was significant in the entire series (p = 0.002) and in the patients who had had a two-tendon tear (p = 0.016), but it was not found to be significant, with the numbers available, in the patients who had had a one-tendon tear (p = 0.5). Fatty muscle degeneration of the infraspinatus progressed in twelve patients (two who had originally had a one-tendon tear, eight who had had a two-tendon tear, and two who had had a three-tendon tear). The progression was significant in the entire series (p = 0.0005) and in the patients who had had a two-tendon tear (p = 0.007), but it was not found to be significant in the patients who had had a one-tendon tear (p = 0.5). In the two-tendon-tear group, fatty muscle degeneration of the infraspinatus progressed in four of the five patients in whom the tendon of the infraspinatus had been intact at the index operation and in the four patients in whom the infraspinatus had not been successfully repaired (Figs. 3-A and 3-B) (Table II). Fatty muscle degeneration of the subscapularis was not found to have progressed significantly (p = 0.147), with the numbers available.
Muscle atrophy of the supraspinatus could be assessed in a standardized fashion on eleven preoperative magnetic resonance images and on all postoperative images. The postoperative cross-sectional area as seen on the Y-shaped image decreased significantly in ten of the eleven patients (p = 0.007); nine of the ten had originally had a two-tendon tear.
Clinical Evaluation
Eleven patients were very satisfied with the result, six were satisfied, two were disappointed, and one was dissatisfied. No patient thought that the shoulder was worse than it had been preoperatively. At the time of follow-up, the mean subjective shoulder value was 75 percent of the value for a normal shoulder. The mean relative Constant score increased from 49 percent preoperatively to 83 percent at the most recent follow-up evaluation (p = 0.0001). Nine patients had a relative Constant score of at least 90 percent (Table III). Patients with a one-tendon tear at the index operation were not found to have a significantly higher mean relative postoperative Constant score than those with a two-tendon tear (92 compared with 80 percent, p = 0.1457). The Constant score correlated well with the subjective shoulder value (r = 0.82, p < 0.0001). Pain decreased significantly (p = 0.0026), with eight patients being pain-free (a Constant score of 15 points) and only one patient having severe pain (a Constant score of 0 points). The ability to carry out activities of daily living was significantly improved (p = 0.0002), with ten patients having no limitations (a Constant score of 10 points), as was functional use of the arm (p = 0.0005), with thirteen patients having no limitations (a Constant score of 10 points). Except for external rotation (p = 0.1805), the ranges of active motion increased significantly (p = 0.0006). The mean abduction strength improved significantly (p = 0.0137), from 1.7 to 3.2 kilograms (Table IV). Except for postoperative abduction strength (4.8 compared with 2.2 kilograms, p = 0.0021), the results for the patients who had been operated on for a one-tendon tear were not found to be significantly better than the results for those who had had a two or three-tendon tear (Table V).
The sixteen patients who had been treated with an acromioplasty were not found to have better clinical results (p = 0.143), with the numbers available, than the four who had not. We also found no difference in the clinical outcome for the nine patients who had had a rupture of the long head of the biceps tendon, a biceps tenotomy, or a biceps tenodesis (p = 0.38). Three patients (Cases 17, 18, and 19) with rupture of the infraspinatus at the index operation were not able to actively externally rotate the treated shoulder at the time of follow-up and had a positive lag sign17 for the supraspinatus and infraspinatus.
The small number of reruptures (four) that were larger than the initial tear did not allow for statistical comparison of preoperative and postoperative parameters in this group. The mean postoperative subjective shoulder value in this group (58 percent) was not found to be significantly different (p = 0.058) from the value for the sixteen patients in whom the rerupture was smaller than the initial tear (79 percent). The four patients with the larger reruptures had a mean relative Constant score of 64 percent postoperatively (45 percent preoperatively) compared with 88 percent postoperatively for the sixteen patients with the smaller reruptures (p = 0.016). All four patients with the larger reruptures had an improvement in the absolute Constant score, although one (Case 17) had only 2 points of improvement. The mean postoperative score for pain in this group (6.3 points) did not improve compared with the mean preoperative score (6.8 points), and the patients with the larger reruptures had more pain postoperatively (6.3 points) than did the patients with the smaller reruptures (12.6 points) (p = 0.03). Except for external rotation, which decreased from 31 degrees preoperatively to 16 degrees postoperatively, the active ranges of motion in the group with the larger reruptures improved compared with the preoperative ranges. Flexion improved from 90 to 124 degrees; abduction, from 91 to 113 degrees; and internal rotation, from 6 to 8.5 points. With the numbers available, no significant difference could be detected between the active ranges of motion of the patients with the larger reruptures and those of the patients with the smaller reruptures (124 compared with 151 degrees for flexion [p = 0.108], 113 compared with 158 degrees for abduction [p = 0.118], 16 compared with 40 degrees for external rotation [p = 0.088], and 8.5 compared with 8.8 points for internal rotation [p = 0.773]). Abduction strength in the four patients who had the larger reruptures was 1.7 kilograms compared with 3.6 kilograms in the patients with the smaller reruptures (p = 0.13). No preoperative or intraoperative risk factors for the development of a rerupture that was larger than the initial tear could be identified.
Correlation Between Preoperative and Postoperative Functional and Imaging Parameters
The postoperative Constant score as well as the postoperative abduction strength (r = -0.75, p = 0.0001, and r = -0.61, p = 0.004, respectively) correlated significantly with the size of the rerupture but not with the size of the preoperative tear (r = -0.24, p = 0.303, and r = -0.08, p = 0.73).
There was no correlation between the postoperative Constant score and fatty muscle degeneration of the supraspinatus (r = -0.27, p = 0.246 preoperatively and r = -0.21, p = 0.378 postoperatively). The Constant score correlated significantly with postoperative fatty muscle degeneration of the infraspinatus (r = -0.55, p = 0.012) and the subscapularis (r = -0.51, p = 0.023) but not with preoperative fatty degeneration of either the infraspinatus (r = -0.11, p = 0.653) or the subscapularis (r = -0.3, p = 0.204).
The degree of osteoarthritis seen on the postoperative radiographs was also significantly correlated with the postoperative Constant score (r = -0.63, p = 0.003) and the size of the rerupture (r = 0.8, p < 0.0001) but not with the size of the preoperative tear (r = 0.38, p = 0.094) or postoperative pain (r = 0.29, p = 0.214).
The postoperative acromiohumeral distance correlated significantly with the postoperative Constant score (r = 0.52, p = 0.02), the size of the rerupture (r = -0.59, p = 0.006), and postoperative fatty degeneration of the infraspinatus (r = -0.61, p = 0.004) but not with postoperative fatty degeneration of the supraspinatus (r = -0.281, p = 0.23).
Return to Work
Preoperatively, nineteen patients were working in their original occupation and one was retired. Eight of the nineteen patients had a strenuous job. Postoperatively, fifteen patients returned to their original occupation, after an average of 3.5 months (range, 0.5 to ten months); two patients who had been performing manual work changed to a less strenuous job; and two patients were receiving a disability pension.
Rerupture or structural failure after rotator cuff repair is a well known and frequently encountered complication1,10,14,16,21,25,32. In studies reported in the literature, reruptures have usually been diagnosed with sonography1,10,16, arthrography4, or arthrography together with computed tomography14. Thomazeau et al.32 and Knudsen et al.21 recently used magnetic resonance imaging for evaluation after rotator cuff repairs. Magnetic resonance imaging is considered to be the primary investigative tool for evaluation after rotator cuff repairs15 and the noninvasive imaging modality of choice for patients with suspected rotator cuff disease24. It is especially helpful for differentiating full and partial-thickness tears of the supraspinatus tendon from intact tendons, a differential diagnosis that may be impossible to accomplish with a physical examination18,24. The sensitivity and specificity of magnetic resonance imaging are higher than those of sonography or arthrography2, especially for full-thickness tears18,24.
In the current study, the postoperative structural integrity of the rotator cuff was assessed with magnetic resonance imaging. All of the twenty reruptures involved the supraspinatus (either alone or with one of the other tendons), as did all of the original ruptures that were treated at the index operation. Lesions of the most superior part of the subscapularis were difficult to identify with use of the magnetic resonance imaging sequences (without arthrography) employed for this study, which may explain why a postoperative lesion of this tendon was detected in only two patients.
The rerupture was smaller than the initial tear in sixteen of the twenty patients. When it was larger, the clinical result was inferior and the patient was less satisfied, although there was still improvement compared with the preoperative state. Three of the larger reruptures had extended posteriorly into the infraspinatus tendon with concomitant functional weakness of this muscle, as documented by the inability to actively externally rotate the arm with the elbow at the side. These observations are in agreement with the concept that a single-tendon lesion of the supraspinatus does not influence the motion pattern of the glenohumeral joint23,33, but the integrity of the infraspinatus is essential for a good clinical result1,34 because it maintains an intact centering mechanism1,3,23,33. The functional importance of the infraspinatus is also demonstrated by the significant correlation of postoperative fatty degeneration of this muscle with the postoperative acromiohumeral distance and the clinical outcome; this correlation was not seen for the supraspinatus.
Glenohumeral osteoarthritis progressed significantly (p = 0.002), especially in patients with a two or three-tendon tear, and was associated with larger reruptures and less favorable clinical outcomes. This was also observed by Petersson28, but Bellumore et al.1 did not find that progression of osteoarthritis influenced the end result. In our series, osteoarthritis did not correlate with postoperative pain.
A key question for patients and surgeons is whether structural failure is identical to clinical failure. Our study documents that failed rotator cuff repairs provide significant improvement compared with the preoperative state. The overall excellent clinical outcome is related not only to successful treatment of pain but also to a gain in overhead motion and strength. Nineteen of the twenty patients felt that they had improvement compared with the preoperative state, and seventeen of the twenty patients were satisfied with the result. On the average, the patients rated the value of the operatively treated shoulder as being 75 percent of the value of a completely normal shoulder. Although the rate of patient satisfaction in our study was lower than that reported in studies of patients with successfully repaired rotator cuffs16, it was at least in the range of the satisfaction rate of 63 to 90 percent reported for patients treated with arthroscopic débridement7,8. Subjectively unsatisfactory results were caused either by postoperative pain (one patient) or by inappropriately high expectations of patients who originally had had a very large tear (two patients).
Four of the twenty patients were completely asymptomatic, and the rerupture would have been impossible to detect on physical examination alone. In fact, in sixteen of the patients, despite the structural failure of the repair, pain relief and restoration of active mobility were comparable with those associated with a postoperatively intact cuff1,10,16. Interestingly, we documented not only a significant decrease in pain and an improvement in overall function but also a significant improvement in measured abduction strength. Whereas quantitative measurements of strength after successful repairs of the rotator cuff have been either reported postoperatively13,16,19,22,35 or compared with preoperative measurements10,20,21,30,32, we know of only a few studies that have documented strength before and after failed repairs21,32, and we do not know of any study of an entire series of such patients. The improvement of measured strength that was found in our series of reruptures was hitherto considered to be consistent with only so-called intact repairs32. The postoperative strength did correlate with the size of the rerupture, as was observed by Rokito et al.30, but our data show that a watertight cuff is not mandatory for recovery of abduction strength of greater than four kilograms, as was previously believed32. Whereas pain relief and functional improvement have also been reported after open29 and arthroscopic7,8,26,37 débridement, we are not aware of any documentation of improvement of measured abduction strength in patients who had a rerupture after open repair of the rotator cuff, and we consider our observation that strength was improved by partial healing to be of importance.
In conclusion, patients with a rerupture after rotator cuff repair still had significant improvement compared with the preoperative state. The rerupture usually was smaller than the original tear, and the structural failures were tolerated well, with good pain relief and functional improvement, including abduction strength. This finding suggests that the potential for structural failure should not be considered to be a formal contraindication to an attempt at rotator cuff repair if optimal functional recovery is the goal of treatment.