The inclusion criteria were patients with (1) full-thickness rotator cuff tears treated by arthroscopic rotator cuff repair, (2) rotator cuff tendon tissue samples harvested at the time of surgery, and (3) availability of either postoperative MRI or computed tomography arthrography (CTA) at 1 year after surgery. Between July 2009 and May 2011, 213 consecutive patients were arthroscopically treated for shoulder problems at our institution. Among them, 146 patients with full-thickness rotator cuff tears were arthroscopically repaired. The remaining 67 patients were excluded, and comprised 49 partial-thickness rotator cuff tears, 1 isolated subscapularis tears, 5 infections, 9 bursitis, and 3 calcific tendinitis. Another 19 patients were excluded from the study since they did not undergo postoperative MRI or CTA at 1-year follow-up. An additional four patients did not undergo tissue biopsy. Finally, 56 patients were included in this retrospective study. Patients were grouped according to the structural integrity of the repaired rotator cuff tendon. Patients with intact tendon integrity according to 1-year follow-up MRI or CTA were classified as the healed group, and patients that showed failure in structural integrity were classified as the retear group.

All arthroscopic surgeries were performed by the same surgeon with patients in the lateral decubitus position under general anesthesia as previously described [2]. Briefly, systemic glenohumeral joint and subacromial exploration were performed, and lesions were managed as necessary. In each case, the rotator cuff tear was carefully evaluated after removing the frayed and atrophied torn end. Anteroposterior (AP) tear size, mediolateral (ML) retraction, number of involved tendons, excursion, presence of subscapularis tear, and gross tendon grade were documented [1,2]. Rotator cuff quality was evaluated using three grades based on gross appearance of the tendon at the time of surgery, based on three criteria: (1) fraying of over half the tendon thickness, (2) delamination, and (3) thinning to less than half the thickness of a normal rotator cuff, that is, less than 6 mm. Each tendon's overall quality was graded as A if none of these criteria were met, B if fraying or delamination were present, and C if both fraying and delamination or thinning were present. If footprint restoration was inadequate, tendon mobilization procedures including superior capsulotomy, coracohumeral ligament release, and medialization of the supraspinatus insertion in the greater tuberosity were performed to improve tendon excursion. Anterior or posterior interval slide was not performed in any patient. The footprint of the greater tuberosity was debrided using a shaver and Arthrocare, and only the unmineralized fibrocartilage layer of the cortical bone was removed. Debridement of bursal tissue, subacromial and distal clavicle osteophytes was minimally performed. Extensive acromioplasty to flatten a hooked or curved acromion was rarely performed. Tissue samples of 3×3 mm in size were harvested from the lateral edge of the torn supraspinatus tendon using a basket forceps after gentle shaving of the torn end. Rotator cuff repair was performed to cover the original footprint using the suture bridge technique whenever possible [1]. Suture anchors were inserted through the accessory portal. Generally, three to five suture anchors were used; one or two anchors for the medial row, and two or three anchors for the lateral row. Medial sutures were placed using either a 5.5-mm Healicoil anchor (Smith & Nephew) or a 5.5-mm Healix anchor (DePuy Synthes Mitek), and lateral sutures were placed using a either Quattro Link knotless anchor (Zimmer Biomet) or Footprint Ultra PK anchor (Smith & Nephew). All knots were tied securely using slippage-proof knots [14].

The shoulder was immobilized for 4 to 6 weeks using an abduction brace (4 weeks for small, medium, and large-sized tears, 6 weeks for massive tears). Shrugging, protraction, retraction of shoulder girdles, intermittent exercise of the elbow, wrist and hand, and external rotation of the arm to neutral with the brace were encouraged as tolerated, usually immediately after surgery. Further passive ROM and active-assisted ROM exercises were allowed after gradual weaning off of the abduction brace from 4 to 6 weeks after surgery. Patients began strengthening exercises after 3 months postoperatively. Light sports activities, such as jogging, were allowed after 3 months, and full return to sports was allowed at 6 to 9 months according to individual recovery.

Healed and retear groups were compared with respect to demographic, histological, clinical, structural, and arthroscopic factors that are thought to be associated with structural integrity of the repaired tendon. For histological assessment, harvested tendon specimens were fixed with 5 mL of neutral buffered 10% formalin in a plastic pathology container. Then, they were embedded in paraffin, cut into 4-µm thick sections, and stained with hematoxylin and eosin. Three sections were prepared per tendon and one of these was randomly selected and examined under a light microscope. Each slide was numbered using randomly generated numbers. After one of the investigators interpreted all of the slides once, the slides were arbitrarily renumbered using a new series of randomly generated numbers.

Tendon samples were graded using a modified semi-quantitative grading scale that assesses degeneration of the rotator cuff tendon [15,16]. The seven parameters included in the grading scale are: (1) fiber structure, (2) fiber arrangement, (3) rounding of the nuclei, (4) regional variations in cellularity, (5) increased vascularity, (6) decreased collagen stainability, and (7) hyalinization. A 4-point scoring system was used; 0 indicates normal; 1, slightly abnormal; 2, moderately abnormal; and 3, markedly abnormal. Overall, the total degeneration score for a given slide could vary from 0 (normal tendon) to 21 (most abnormal appearance detectable). All examinations were performed by two blinded investigators and two fellowship trained orthopedic surgeons. For each slide, the entire area was examined and the worst area was selected for comparisons between slides. The results were assessed for intra-observer and inter-observer reliability by reassessing the slides 1 week after the first examination. After assessing inter- and intra-observer variation, we used the results of the second examinations in this study.

Preoperative clinical outcomes were evaluated by the same orthopedic team. Prior to surgery, each patient completed a questionnaire that included standardized outcome assessments. Clinical factors included visual analog scale (VAS) pain score, ROM, strength, and preoperative functional scores (American Shoulder and Elbow Surgeons system, the Constant system, University of California Los Angeles system, Disabilities of Arm, Shoulder and Hand system, Simple Shoulder Test, and Shoulder Pain And Disability Index system). Clinical factors were assessed according to (1) pain, (2) ROM, (3) muscle strength, and (4) functional scores. VAS pain score was used to assess pain at rest, on motion and at night. Patients were asked to use a 10-cm scale marked from “no pain” to “unbearable pain.” The average VAS pain scores were calculated in both groups and compared preoperatively. ROM was measured with a goniometer in active forward flexion, abduction, external rotation with the arm at the side, and internal rotation. Internal rotation was measured using vertebral levels, and these were translated into numbers from 1 for the buttocks to 17 for T2 level. Flexion motor power, external rotation motor power, and internal rotation motor power were measured using a hand-held electronic scale (CHS; CAS).

To evaluate structural integrity, MRI (Achieva 3.0-T; Philips Medical System) with a dedicated shoulder coil or CTA (Ingenuity CT scanner; Philips) were performed at a minimum of 9 months after surgery. The structural integrity was evaluated using Sugaya’s classification for patients with MRI [17], or a modified Boileau’s grading system for patients with CTA [2,8]. In Sugaya’s classification, type I, II, and III were considered structurally healed tendons and type IV and V were considered retears. In the modified Boileau’s grading system, healed and incompletely healed were considered “healed,” while retear and new tear were considered “retear” [2,8]. All images were reviewed by a radiologist and an orthopedic surgeon. When there were discrepancies in reading, the worse evaluation was recorded.

Other structural factors included fatty infiltration and muscle atrophy. For the qualitative assessment of fatty infiltration, the Goutallier classification modified by Fuchs was used [18]. The classification has five stages, ranging from stage 0 (normal muscle) to stage 4 (more fat than muscle): stage 0, normal muscle without fat; stage 1, some fatty streaks within the muscle; stage 2, less fat than muscle; stage 3, same amount of fat and muscle within the muscle; and stage 4, more fat than muscle within the muscle. Muscular atrophy of the supraspinatus muscle was qualitatively graded using a modified Tangent sign [18]. The tangent line was drawn between superior margin of the scapular spine and coracoid. The grading system has three stages: grade 1, supraspinatus muscle crossing over the tangent line; grade 2, supraspinatus muscle meets the tangent line; and grade 3, most superior portion of the supraspinatus muscle is under the tangent line. The occupation ratio (OR) described by Thormazeau et al. [19] was used for grading the quantitative measurement of muscular atrophy.

Arthroscopic factors included rotator cuff tendon tear size, tendon retraction length, and tendon morphology. The extent of tear size and retraction was determined during operation under direct arthroscopic visualization after minimal debridement of the torn tendon edge. Macroscopic tendon quality was assessed based on three criteria: (1) fraying of over half the tendon thickness, (2) delamination, and (3) thinning to less than half the thickness of the normal rotator cuff, that is less than 6 mm. Gross tendon quality was graded as A if none of these criteria were met, B if either fraying or delamination was identified, and C if both fraying and delamination or thinning with or without the other two criteria were identified [2].

All analyses were performed using SPSS ver. 13.0 (SPSS Inc.). The mean values were compared using Student t-test or the Mann-Whitney U-test for continuous variables (age, symptom duration, symptom aggravation, total degeneration score, clinical scores, and tear size), and Pearson’s chi-square test or Fisher’s exact test for categorical variables (sex, fatty infiltration, tangent, occupation, tendon grading, and histological factors) to determine differences according to postoperative structural integrity. The paired t-test or Wilcoxon signed-rank test was used to compare preoperative functional scores. To validate tendon histopathology grading in this study, we assessed intra-observer and inter-observer variation by interclass correlation coefficient (ICC). Kappa statics were used to assess agreement between slide scores. In addition, multivariate logistic regression analysis was used to identify independent variables affecting outcomes using the stepwise forward conditional method. A P-value of 0.05 was considered significant.

Between July 2009 and May 2011, 56 patients were included in this study and retear was found in 12 patients (21.4%). Thus, 44 patients were grouped as the healed group, and the other 12 patients were grouped as the retear group. There were no significant preoperative differences in baseline demographic factors including gender, symptom duration, aggravation and dominance between the two groups (Table 1).

Histological assessment with a modified semi-quantitative grading scale showed high inter- and intra-observer reliability (Table 2). The ICCs of the inter-observer and intra-observer reliability for the total degeneration score were 0.899 (P<0.001) and 0.777 (P=0.001), respectively.

The mean scores in the healed and the retear group were 13.93±2.03 and 14.08±2.23 (P=0.960), respectively. No histological variables were significantly different.

Mean fatty infiltration of the supraspinatus in the healed and the retear groups were 1.81±0.92 and 2.41±0.99 (P=0.055), respectively. Mean fatty infiltration of the infraspinatus in the healed and the retear groups were 1.30±0.79 and 1.51±0.52 (P=0.404), respectively. Mean fatty infiltration of the teres minor in the healed and the retear groups were 0.79 ±0.70 and 0.75±0.62 (P=0.840), respectively. Mean values of modified tangent sign in the healed and the retear groups were 1.30±0.46 and 1.67±0.88 (P=0.052), respectively. Mean values of occupation ratio grade in the healed and the retear groups were 1.70±0.66 and 2.08±0.79 (P=0.100), respectively. Although there were no significant differences among the two groups, mean values of structural factors in the healed group were lower for all variables.

Mean tear size of AP dimension in the healed and the retear groups were 24.30±12.35 mm and 36.42±25.23 mm, respectively (P=0.026). Mean retraction sizes measured in ML were 16.02±7.58 mm and 22.33±13.36 mm (P=0.037), respectively. The mean values of tendon grade in the healed and the retear group were 1.82±0.582 and 2.08±0.79 (P=0.199), respectively.

In the univariate analysis, the retear group showed significant differences with higher preoperative VAS pain score at rest (P=0.046) and larger tear size in both AP (P=0.026) and ML (P=0.037) dimensions compared to the healed group. Multivariable analysis showed that larger AP tear size was associated with postoperative structural integrity. The odds of retear rose by 4.2% for every 1mm increase in AP tear size (Exp(B): 1.042; 95% confidence interval: 1.004–1.082; P=0.032). AP tear size assessed during surgery can be used to estimate retear rate using the following logistic regression: logit(p)=β1×(AP tear size)+β2×(pre-VAS at rest)+β3(ML tear size).