Factors associated with healing failure after early repair of acute, trauma-related rotator cuff tears

Background: Healing failure after rotator cuff repair is a challenging problem. Acute, trauma-related tears are considered a separate entity and are often treated surgically. The aim of this study was to identify factors associated with healing failure in previously asymptomatic patients with trauma-related rotator cuff tears treated with early arthroscopic repair. Methods: This study included 62 consecutively recruited patients (23% women; median age, 61 years; age range, 42-75 years) with acute symptoms in a previously asymptomatic shoulder and a magnetic resonance imaging–veriﬁed full-thickness rotator cuff tear after shoulder trauma. All patients were offered, and underwent, early arthroscopic repair, during which a biopsy specimen was harvested from the supraspinatus tendon and analyzed for signs of degeneration. Of the patients, 57 (92%) completed 1-year follow-up and underwent assessment of repair integrity on magnetic resonance images according to the Sugaya classiﬁcation. Risk factors for healing failure were investigated using a causal-relation diagram where age, body mass index, tendon degeneration (Bonar score), diabetes mellitus, fatty inﬁltration (FI), sex, smoking, tear location regarding integrity of the rotator cable, and tear size (number of ruptured tendons and tendon retraction) were included and analyzed. Results: Healing failure at 1 year was identiﬁed in 37% of patients (n ¼ 21). A high degree of FI of the supraspinatus muscle ( P ¼ .01), a tear location including disruption of rotator cable integrity ( P ¼ .01), and old age ( P ¼ .03) were associated with healing failure. Tendon degeneration as determined by histopathology was not associated with healing failure at 1-year follow-up ( P ¼ .63). Conclusion: Older age, increased FI of the supraspinatus muscle, and a tear including disruption of the rotator cable increased the risk of healing failure after early arthroscopic repair in patients with trauma-related full-thickness rotator cuff tears.

Patients with trauma-related rotator cuff tears (TRCTs) without previous shoulder complaints are typically recommended to undergo early repair. 22,48,60The treatment goals of TRCT repair are tendon healing and restoration of anatomy and biomechanics leading to regained normal shoulder function.Several prognostic factors for repair healing and risk factors for nonhealing have been studied during the past few decades, but to our knowledge, investigations of solely trauma-related tears have not been published. 32,37,66Among factors independently associated with lower rates of repair healing, 3 different variables for cuff muscle degeneration have been reported: high grade of fatty infiltration (FI) of the supraspinatus 42,63 ; high grade of FI of the infraspinatus 13,14,31,36,52,63 ; and elevation of the combined, calculated Global Fatty Degeneration Index. 12,21,51n addition, large tear size, 12,15,27,36,42,43,52,55,56 older patient age, [15][16][17]27,42,43,52,61 hyperlipidemia, 20,26,36 increased body mass index, 17,36 muscle atrophy, 36 and low bone mineral density 14 have been reported as independent risk factors for healing failure. A large proporon of TRCTs in previously asymptomatic patients have recently been shown to have histologically verified degenerative changes. 1 Identifying risk factors for healing failure is required to understand current limitations in achieving a successful repair and, furthermore, to guide both patients and surgeons in treatment decision making.
We hypothesized that tear location and degenerative changes might predict healing failure.These relationships have not been investigated or reported in similar cohorts.The aim of this study was therefore to identify risk factors associated with healing failure in patients with TRCTs treated with arthroscopic repair within 6 weeks of trauma.

Patient selection
All data were collected from patients in a prospective clinical study at Helsingborg Hospital.The study enrolled 62 consecutive patients, aged between 42 and 75 years, with pain and limited range of motion following shoulder trauma and with normal preinjury function between November 2010 and March 2014.Details of the study have recently been published. 2In brief, the inclusion criteria were a full-thickness rotator cuff tear verified by magnetic resonance imaging (MRI), no previous shoulder complaints, and willingness to undergo surgical treatment within 6 weeks of trauma.The exclusion criteria were previous surgery on the affected shoulder; history of shoulder dysfunction; and presence of rheumatoid arthritis, glenohumeral osteoarthritis, nerve injury, or severe comorbidity (American Society of Anesthesiologists class !4).Eligible patients were examined by a dedicated physiotherapist in the orthopedic department at a median of 12 days (interquartile range [IQR], 10-16 days) after injury when data on smoking habits, body weight, and height were collected.MRI of the affected shoulder was performed within a median of 19 days (IQR, 15-24 days) after injury.Eligible patients with a full-thickness rotator cuff tear were offered early arthroscopic repair, which was performed by a single-row technique within a median of 30 days (IQR, 25-37 days) after injury.All surgical procedures were performed by 1 of 3 experienced orthopedic surgeons at Helsingborg Hospital.After visualization, the type of tear was categorized and the tear size was measured in both the coronal plane (retraction in millimeters) and the sagittal plane (number of tendons involved) with a calibrated probe.Measurements and concomitant pathology were recorded in a predefined surgical protocol (Table I).Isolated crescent tears of the supraspinatus and crescent tears involving the most anterior part of the infraspinatus were, together with single-tendon subscapularis tears, classified as rotator cable (RC)-stable tears.All other tears involving the anterior and/or posterior cable structures were considered RC-unstable tears (Fig. 1). 11,23,38,47Biceps pathology was treated with tenotomy or tenodesis according to the surgeon's preference.All subscapularis tears were repaired with Fastin RC 5.0-mm anchors (DePuy Synthes Sports Medicine [Mitek], Raynham, MA, USA), whereas for infraspinatus and supraspinatus tears, triple-loaded Healix 4.5-and 5.5-mm anchors (DePuy Synthes Sports Medicine) were used.Subacromial decompression with a burr was performed in all patients except those undergoing isolated subscapularis repair.After surgery, all patients followed a standardized rehabilitation protocol including use of a sling for 4-6 weeks (dependent on infraspinatus involvement, described in detail in a previous report 2 ).A biopsy specimen was harvested from the most lateral part of the supraspinatus tendon stump before any use of radiofrequency instruments.Tissue material was fixed in neutral buffered formalin in a standard fashion and sent to the Department of Pathology at Lund University for dehydration and paraffin embedding.Sections of 2 mm were obtained; stained with hematoxylin-eosin, alcian blue, and Masson trichrome; and finally assessed for tendon degeneration according to the Bonar score. 18The median Bonar score of patients with trauma-related tears was 10.5, not statistically significantly different from that of a control group comprising patients with chronic tears.The complete results of the histopathologic analyses have recently been published. 1

Magnetic resonance imaging
A standardized protocol with 7 diverse sequences was applied, and all MRI examinations were conducted with the same 1.5-T scanner (Siemens Medical Systems, Erlangen, Germany).All preoperative images were interpreted and classified by a single senior radiologist.A full-thickness rotator cuff tear was defined as tendon discontinuity or increased signal on T2-weighted images, isointense compared with fluid, extending from the articular side to the bursal side of the tendon. 29The integrity of the different cuff tendons was documented, together with the grade of FI of the subscapularis, supraspinatus, and infraspinatus muscles according to the 3 stages of the Fuchs scale 19 : stage 1, no to some fatty streaks; stage 2, pronounced FI but fat area smaller than muscle area; and stage 3, fat area equal to or larger than muscle area.The Global Fatty Degeneration Index was calculated for each patient using the mean FI value of the 3 rotator cuff muscles as described by Goutallier et al. 21Intraobserver reliability of the radiologist's grading of FI was tested by Cohen k calculation with the 95% confidence interval by reassessment of one-third of the images, randomly selected, 2 months after initial assessment (Table II).As suggested previously, k values between 0.41 and 0.60 were considered moderately reliable; between 0.61 and 0.80, substantially reliable; and >0.80, almost perfectly reliable. 39n 57 patients (92%), a follow-up MRI scan was performed at a median of 63 weeks (IQR, 58-69 weeks) after surgery to evaluate the structural integrity of the repair according to the Sugaya classification. 59Loss to follow-up has been reported in detail. 2 Structural integrity was defined and dichotomized based on the Sugaya grade as healing failure (Sugaya grade 4 or 5) and intact repair (Sugaya grade 1-3).PAGE y# "'Page: '#''" 25,33 Healing failure was denoted by !1 repaired tendon showing Sugaya grade 4 or 5.A second senior radiologist reassessed the follow-up MRI scans blinded to clinical findings.

Statistical analysis
Descriptive statistics were used to define distributions of continuous variables, and frequencies, and proportions of categorical variables.Variables previously suggested to influence healing [12][13][14][15]17,20,26,27,31,42,43,51,52,55,56,61,63 were included in a directed acyclic graph (DAG) model (Fig. 2) to assess causal relations. PAGE y# "'Page: '#''" 58,62 Compliancewith the rehabilitation protocol, genetic factors, and bone mineral density were included as uncontrolled variables (light gray in Fig. 2), whereas surgeon's experience and rehabilitation protocol (white in Fig. 2) were fixed variables and thus labeled ''adjusted.''All blue factors in the DAG were generated by the model to have potential impact on the outcome (''ancestor of outcome'' in Fig. 2) and were compared between the intact repair group and the healing failure group using the Mann-Whitney U test and c 2 test.The Fisher exact test was used in place of the c 2 test in instances of a 2-by-2 table with frequency value 5 in at least 1 cell.To investigate intraobserver and interobserver reliability between MRI examiners, k statistics were used.SPSS software (version 27; IBM, Armonk, NY, USA) was used for all statistical analysis.P < .05 was considered significant.

Results
Patient demographic characteristics and preoperative MRI and intraoperative findings are shown in Table I.Of the 57 patients with MRI scans at follow-up, 36 (63%) had intact repairs and 21 (37%) had healing failure according to the Sugaya classification.Interobserver reliability, calculated by k analysis, showed a k value of 0.88 (95% confidence interval, 0.75-1.00),which indicated almost perfect reliability. 39In the univariate analyses, tendon degeneration as determined by histopathology was not associated with healing failure at 1-year follow-up (P ¼ .63).Factors significantly associated with healing failure were a tear location including disruption of RC integrity (P ¼ .01), a   higher degree of FI of the supraspinatus muscle (P ¼ .01),and older age (P ¼ .03)(Table III).

Discussion
This prospective study demonstrated that older age, increased FI of the supraspinatus muscle, and disruption of the RC attachments were associated with healing failure after early arthroscopic repair.The findings are based on a well-defined cohort of acute, trauma-related cuff tears.Only 1 previous report has suggested a disrupted RC as a negative predictor for healing, 11 whereas age and FI have been more commonly described as negative predictors for healing of degenerative tears. 13,42,51,52,63,66It is interesting to note that degeneration of the supraspinatus tendon as determined by histopathology was not associated with healing failure in this study.The RC, first described by Burkhart et al, 7 was suggested to be critical for mechanical properties such as force transmission within the rotator cuff.Disruption of its anterior attachment has been described to be associated with fatty muscle infiltration, tear progression, 49 and failed repair of degenerative tears. 11As opposed to tear location, tear sizeddetermined by the number of involved tendons, as well as tear retractiondwas not found to be a risk factor for healing failure in our study.This may suggest that RC involvement is more important than the actual tear size in millimeters.Restoration of anatomy and secure fixation of the RC anchor sites seem important to shield mechanical stress, tallying with the conclusion in a biomechanical study by Nguyen et al. 50Thus, disruption of the RC should possibly be targeted in specifically designed future trials.Our results may be interpreted to support rotator cuff repairs with augmentation to enhance healing by securing the construct.The use of biceps superior capsular reconstruction is one option with promising results. 3,5,9,10s previously reported, 37% of our patients with an assumingly fresh cuff tear were identified with healing failure 1 year after surgery, despite early arthroscopic repair. 2 This is comparable to the 24%-35% healing failure rates reported in other studies on similar trauma-related tear cohorts. 6,24,30,65Still, a failure rate as low as 6.5% has been presented in a cohort with smaller tear sizes (median, 10 mm) and, in most cases, with an intact RC. 54 In degenerative tears, an association between age and cuff integrity after repair has been well documented. 15,16,42,52However, whether age is an independent factor or a confounding factor is not clear, as age is closely related to other important features such as tendon degeneration, osteoporosis, and FI. 4,34,35Multivariate analyses performed in studies with larger cohorts of patients with degenerative tears have suggested that age, tear size, follow-up duration, and FI of the supraspinatus are independent risk factors for healing failure. 42,61roxies for tendon degeneration, such as surgeon-rated tissue quality, 28,41,64 as well as inflammatory and matrix remodeling biomarkers, 45 have been reported as potential risk factors for healing failure.Surprisingly, histopathologic degenerative changes of the involved supraspinatus tendons, recently shown to be as common in this traumarelated tear cohort as in chronic nontraumatic tears, 1 were not associated with healing failure in this study.This agrees with a report by Sethi et al, 57 who similarly failed to find a relationship between the Bonar score, gross appearance of the rotator cuff, and repair healing.One possible explanation is that tendon degeneration precedes more devastating muscle changes such as FI.Although smoking has a wellknown negative influence on different tissue qualities, including increased apoptosis and histopathologic changes in rotator cuff tissue, 44 we failed to find a relationship to repair failure in this study.This is most likely a result of an underpowered analysis because the cohort only included 6 smokers (10%).
In our study, FI was analyzed as a global index, as well as in individual muscles separately.Surprisingly, only 79% of the individuals had a normal index of global FI preoperatively despite the fact that our cohort comprised traumarelated tears in individuals without any history of shoulder dysfunction.It appears unlikely that the observed muscle changes would have occurred during the relatively short period between the traumatic event and the examination, and an asymptomatic preceding degenerative process over time prior to the traumatic event seems probable.In agreement with the authors of several other studies on degenerative tears, 12,21,40,42,63 we identified a relationship between increased FI of the supraspinatus and healing failure in our cohort.This association is possibly of high importance for healing failure in repaired supraspinatus tears; an odds ratio of 9.3 for healing failure was suggested in a meta-analysis. 53In theory, FI may directly affect the structural outcome after rotator cuff repair by reduced muscle strength and secondary superior migration of the humeral head, leading to impingement and repair damage as suggested by Goutallier et al. 21he use of a DAG model (Fig. 2) to avoid bias, where included variables were collected from previous studies and arranged from a clinical perspective, is a strength of this study.However, this resulted in a large number of potential variables that could confound or mediate the causal effect on the outcome.The relationships between FI, muscle atrophy, tear location, size, and age are drawn in the DAG model, yet the quantity of the causal effect of different variables is not known.The selection of variables and their potential relationship was based on the systematic review and meta-analysis by Zhao et al 8,12,26,46,63,66 from 2021, together with other recent studies on healing after rotator cuff repair. 8,12,26,46,63,66here are limitations to this study.First, the number of patients is relatively small, and the results should be interpreted with caution because the analyses could have been underpowered.Second, although we used a clear definition of acute, trauma-related tears, it is possible that some of our patients had asymptomatic tears before the traumatic event, which could explain the FI observed on the preoperative MRI scans in 20% of the patients.Third, a few potential risk factors were not studied, such as reduced bone mineral density and genetic factors, and their relationship to healing failure remains to be determined in studies with larger samples.Finally, the process of finding a  patient suitable for surgical treatment is always susceptible to selection bias.A specific study strength is that this is the first study on a selected cohort with trauma-related tears analyzed in relation to repair integrity.The patients were prospectively studied according to a predetermined protocol, which improves representativeness, and we used standardized methodology with blinded examiners performing radiologic and histologic assessments.
The clinical implication of our findings is that patients with trauma-related tears should be informed about the risk of healing failure related to tear location, older age, and FI.Specific attention should be focused on RC disruption and the possibility of repair augmentation.

Conclusion
Tear location including disruption of rotator cable integrity, older age, and increased FI of the supraspinatus muscle increased the risk of healing failure after early arthroscopic repair in patients with traumarelated full-thickness rotator cuff tears.These factors are therefore recommended to be considered before surgery.A relationship between histopathologic degenerative changes within the supraspinatus tendon and repair integrity was not found.

Figure 1
Figure 1 Example of crescent tear involving both supraspinatus and infraspinatus with disruption of posterior insertion of rotator cable (lateral view in right shoulder).

Figure 2
Figure 2 Directed acyclic graph.) As determined by Bonar score.BMI, body mass index; rehab, rehabilitation.

Table I
Patient demographic characteristics and preoperative and intraoperative assessment findings