Risk factors for retear of large/massive rotator cuff tears after arthroscopic surgery: an analysis of tearing patterns

Between April 2005 and August 2013, 150 patients with cuff tears defined as large or massive [6] underwent ARCR in our institution. The inclusion criteria were (1) individuals who had large or massive rotator cuff tears that repaired completely during surgery, (2) those who were available for evaluation of function and magnetic resonance imaging (MRI) preoperatively and at a minimum of 1 year after surgery, and (3) those who underwent an appointed postoperative rehabilitation program. The exclusion criteria were (1) individuals with advanced glenohumeral arthritis or fractures around the shoulder, (2) those who underwent open repair, partial repair, revision surgeries, or any previous shoulder surgery, (3) those who had MRI film without the “Scapula - Y” view on the sagittal-oblique plane, (4) those who refused to undergo postoperative clinical assessment and MRI, and (5) those who had preoperative stiffness that showed less than 100° in passive elevation or 10° in external rotation [10]. Consequently, 102 patients were enrolled in this study.

According to the arthroscopic and MRI findings, the 102 patients were divided into three groups based on the tendon location: anterosuperior tears [11] in the subscapularis and the supraspinatus, in which the tear extended from the lesser tuberosity to the superior facet (N = 59, group AS); posteosuperior tears in the supraspinatus and the infraspinatus/teres minor, in which the tear extended from the superior facet to the middle or inferior facet (N = 21, group PS); and anteroposterior-extending tears, in which the tear extended from the lesser tuberosity to the middle or inferior facet (N = 22, group APE). When large or massive tears were evaluated according to the classification of DeOrio and Cofield [12], there were 56 patients with a large tear (94.9%) and 3 patients with a massive tear (5.1%) in Group AS, 19 patients with a large tear (90.5%) and 3 patients with a massive tear (9.5%) in group PS, and 15 patients with a large tear (58.2%) and 7 patients with massive tears (31.8%) in group APE. There was no statistical difference in demographic data among the three groups, except for the incidence of hypertension and distribution of massive tears. Details of the patients’ characteristics are shown in Table 1.

Arthroscopic surgery was indicated when successful non-operative treatment, such as anti-inflammatory medications, physical therapy, subacromial or glenohumeral injections of corticosteroids or hyaluronic acid, or activity modification, was not achieved within 3 months of the first visit.

ARCR was conducted with the patient in the beach chair position under general anesthesia. At first, a glenohumeral examination was performed through a posterior portal and then transferred to the subacromial bursa. After making a lateral portal, we identified the ruptured tendon edge and evaluated its flexibility by grasping the tendon and reducing the edge to the original footprint. Capsular release was conducted from the anterior, anterolateral, or posterolateral portal; if needed, tenotomy of the long head biceps was performed. The method of cuff repair was selected based on the operative findings, tendon mobility, and tear condition with a single-row, double-row, or suture bridge technique (Table 1).

Postoperatively, the patient’s arm was fixed into a sling with an abduction pillow. Passive range of motion (ROM) exercises of the shoulder were conducted 4 days after surgery. Active ROM exercises and isometric exercise were started 6 weeks after surgery, and isotonic muscle strengthening exercises began 12 weeks after surgery.

Functional outcome was evaluated preoperatively and postoperatively. The visual analog scale was used to measure pain (rest, night, and motion), the range of active motion was measured with a goniometer, muscle strength was measured with a handheld dynamometer (Micro FET2, Hoggan Health Industry, West Jordan, UT, USA), Japanese Orthopedic Association (JOA) score, and the University of California, Los Angeles (UCLA) score. An independent physiotherapist who was blinded to this study performed physical tests.

Acromiohumeral distance was evaluated, using the Oizumi classification [13], on plain radiographs that were taken with the patients standing and their arm held in a neutral position.

Tear length and width were measured on MRI using the protocol of Davidson et al. [14] FD of the supraspinatus, the infraspinatus/teres minor, and the subscapularis were evaluated on the most lateral oblique sagittal T2-weighted MRI with the scapular body (the “Y-view”) [15, 16], using both Goutallier classification system and ImageJ [14]. The infraspinatus and teres minor were combined into a single measurement, because their borderline was not always clearly confirmed [17]. Muscle atrophy (MA) was evaluated using the relative ratio of the cross-sectional area of the subscapularis, supraspinatus, and infraspinatus/teres minor muscle belly to that of the supraspinatus fossa. For this measurement, we used ImageJ using the protocol of Nakamura et al. [18]

Retear of the rotator cuff was evaluated using Sugaya’s classification [19]: type I, sufficient thickness and evenly low intensity; type II, sufficient thickness and heterogeneous high intensity; type III, repaired cuff tear that kept its continuity but had insufficient thickness; type IV, minor discontinuity and the torn area was minimal in the sagittal plane; and type V, major discontinuity and torn area spread in the sagittal plane. Patients with types IV and V were admitted with postoperative retear [20]. An experienced, orthopedics-trained radiologist who was blinded to the study reviewed these images.

The statistical analysis was performed with JMP11 software (SAS, Cary, NC, USA). The Kruskal-Wallis test or χ ² test was used to compare the continuous or nominal variables in demographics and functional and structural outcomes among the three groups. A Wilcoxon test was used for comparing the preoperative and postoperative functional outcomes in each group. Spearman’s ρ was calculated to observe the nonparametric correlation of structural outcomes and clinical outcomes. The correlation between the data evaluated with Goutallier’s classification and ImageJ was examined with Spearman’s correlation coefficient. For identifying the risk factors for retear after surgery, univariate analysis was first performed in each group; then, multivariate logistic regression was performed using a step wise manner. A receiver-operating curve was calculated to detect the cut-off value when significance was noted in the multivariate analysis. The level of significance was defined for all calculations as P < .05. Data are expressed as a mean value with standard deviation.

Preoperative JOA scores significantly improved from 57.9 ± 19.9 points preoperatively to 87.4 ± 10.0 points postoperatively in group AS, from 56.9 ± 15.4 points to 89.9 ± 6.6 points in group PS, and from 61.7 ± 7.5 points to 83.0 ± 11.4 points in group APE. Consistently, UCLA scores significantly improved from 18.2 ± 5.0 points preoperatively to 28.3 ± 7.2 points postoperatively in Group AS, from 17 ± 4.7 points to 29.4 ± 3.8 points in group PS, and from 15.9 ± 4.0 points to 28.3 ± 5.9 points in group APE. There were no significant differences of postoperative JOA/UCLA scores among the groups. The clinical outcomes scores are shown in Table 2.

In the Oizumi classification, Class 0 was observed in 19, 6, and 3 patients; Class I in 30, 7, and 8 patients; Class II in 8, 7, and 4 patients, and Class III in 2, 1, and 5 patients in Group AS, PS, and APE, respectively. Only two patients in group APE had Class IV.

The average retraction of the torn tendon was 29.1 ± 6.2 mm in group AS, 31.2 ± 10.4 mm in group PS, and 37.1 ± 8.7 mm in group APE. The extent of the retraction was significantly larger in group PS and APE than group AS (P = 0.02). The average width of the torn tendon was 34.7 ± 66.0 mm in group AS, 37.2 ± 10.4 mm in group PS, and 45.7 ± 89.8 mm in group APE. The extent of the width was significantly larger in group PS and APE than in group AS (P = 0.002).

The average MA in group AS was 246.5 ± 83.5% in the subscapularis, 76.2 ± 19.8% in the supraspinatus, and 220 ± 51.9% in the infraspinatus/teres minor. For group PS, MA was 220.3 ± 67.4% in the subscapularis, 78.9 ± 18.6% in the supraspinatus, and 183.4 ± 38.5% in the infraspinatus/teres minor; for group APE, MA was 252.1 ± 82.4% in the subscapularis, 71.3 ± 16% in the supraspinatus, and 187.4 ± 54.8% in the infraspinatus/teres minor.

The average FD, measured with Image J, for group AS was 5.35 ± 8.25% in the subscapularis, 7.84 ± 10.24% in the supraspinatus, and 3.35 ± 4.92% in the infraspinatus/teres minor; for group PS, FD was 3.5 ± 5.2% in the subscapularis, 11.42 ± 10.15% in the supraspinatus, and 8.52 ± 9.23% in the infraspinatus/teres minor; and for group APE, FD was 6.0 ± 7.23% in the subscapularis, 12.2 ± 9.7% in the supraspinatus, and 6.05 ± 5.2% in the infraspinatus/teres minor.

A low-grade Goutallier stage (stages 0 to 2) was seen in over 80% patients in all three groups. A high-grade Goutallier stage (stages 3 and 4) in group AS was seen in the subscapularis of two patients, supraspinatus of nine patients, and infraspinatus/teres minor of two patients; in Group PS, in the subscapularis of no patients, supraspinatus of three patients, and infraspinatus/teres minor of three patients. In Group APE, a high-grade Goutallier stage was seen in the subscapularis of two patients, supraspinatus of five patients, and infraspinatus/teres minor of three patients. The global fatty degeneration index (GFDI) was 1.02 ± 0.62 in Group AS, 1.36 ± 0.5 in group PS, and 1.29 ± 0.5 in group APE. There was no significant difference among the three groups in GFDI (Table 3).

Postoperative retear (Sugaya types IV and V) was noted in 26 of 102 patients (25.5%) in this series: 10 patients in group AS (16.9%), 9 patients in group PS (42.9%), and 7 patients in group APE (31.8%). The retear rate was significantly higher in group PS than in groups AS and APE (P = 0.02) (Table 4).

Suture bridge technique was performed in 71 patients: 40 in group AS;14 in group PS; 17 in group APE. Postoperative retear occurred in 16 patients (22.5%): 6 in group AS (15.0%); 4 in group PS (28.6%); 6 in group APE (35.3%).

Single-row technique was performed in 20 patients: 14 in group AS; 4 in group PS; 2 in group APE. Postoperative retear occurred in 8 patients (40.0%): 4 in group AS (28.6%); 3 in group PS (75.0%); 1 in group APE (50.0%).

Double-row technique was performed in 11 patients: 5 in group AS; 3 in group PS; 3 in group APE. Postoperative retear occurred in 2 patients (18.2%): none in group AS (0.0%); 2 in group PS (66.7%); none in group APE (0.0%). Details are shown in Table 5.

First, various parameters were evaluated to determine the risk factors for retear after surgery, using univariate analysis. Retraction (P = 0.039), width (P = 0.0023), FD of the supraspinatus (P = 0.0043), the Goutallier classification of the supraspinatus (P = 0.001), and GFDI (P = 0.008) were significant risk factors in group AS: Preoperative active external rotation range (P = 0.001), preoperative muscle strength of flexion (P = 0.02), and FD of the infraspinatus/teres minor (P = 0.048) were evaluated using ImageJ in group PS. FD of the supraspinatus (P = 0.002) and the infraspinatus/teres minor (P = 0.0074) were evaluated using ImageJ and GFDI (P = 0.0123) in group APE. The Goutallier stage of the infraspinatus in group PS and APE was not a significant risk factor, but it tended to have statistical significance (Table 6).

Next, multivariate analysis using stepwise methods was performed. Preoperative external rotation range was the only risk factor for postoperative retear in groups PS and APE (P = 0.014 and P = 0.016, respectively). For the prediction of postoperative retear, receiver operating characteristic (ROC) curve analysis demonstrated that the cut-off value in the preoperative external rotation range was 25°, showing that the retear risk increased 2.12-fold as the preoperative external rotation range decreased by 5° (Fig. 1).

Since multivariate analysis showed that active external rotation range (AERR) is a unique risk factor for postoperative retear in groups PS and APE, we further evaluated the correlation between AERR and its related variables in these groups. There was statistical significance between AERR and FD, using ImageJ (r = −0.36, P = 0.04). External rotation strength (ERS) was not significant, but it showed a trend (P = 0.06). For FD of the infraspinatus, statistical significance was seen between the evaluation methods, using the Goutallier classification and ImageJ (r = 0.82, P < 0.0001) (Table 7).