Tendon stump change as a risk factor for structural failure of arthroscopic rotator cuff repair

Abstract

Background

This study aimed to identify risk factors for retear after arthroscopic rotator cuff repair (ARCR) using the suture bridge method and to assess the relation between preoperative and postoperative tendon condition and retear.

Methods

We retrospectively analyzed patients who underwent ARCR for a medium tear. Potential risk factors, including tear width, anteroposterior and medial-lateral tear width, atrophy, fatty degeneration, and condition of the tendon stump (Ishitani's classification and rotator cuff/deltoid muscle [C/D] ratio), were identified on magnetic resonance preoperatively. Magnetic resonance imaging was performed regularly after surgery, and the C/D ratio in the sutured cuff was evaluated. Shoulders that were Sugaya type I-III were classified as the repair group and those that were type IV or V as the retear group. Risk factors for retear were analyzed by univariable and multivariable analyses.

Results

We analyzed 175 patients in 183 shoulders. The overall retear rate was 13.1%. Multivariable analysis revealed medial-lateral width (odds ratio: 0.88; 95% confidence interval: 0.82-0.95; P < .001) and C/D ratio (odds ratio: 0.30; 95% confidence interval: 0.16-0.58; P < .001) as independent risk factors for retear. At 3 weeks postoperatively, the C/D ratio was significantly higher in the retear group (1.70 ± 1.47 vs. 0.84 ± 0.72; P = .002).

Conclusions

The C/D ratio and tear width were risk factors for retear after ARCR using the suture bridge method. The C/D ratio was higher in retear cases both preoperatively and postoperatively.

Arthroscopic rotator cuff repair (ARCR) has become a standard method for repair of rotator cuff tears but has a retear rate of 8%-42%.^3,9,13,14,25 Retears have been associated with various risk factors.^3,6,13,25,33 Although the suture bridge (SB) repair method proposed by Park et al²⁴ is considered to be one of the secure methods, the success of initial fixation is greatly influenced by the quality of the rotator cuff.²⁴ Cho et al classified the retear pattern based on the portion of the tear.⁸

Ishitani et al investigated the signal intensity of the rotator cuff tendon stump on magnetic resonance imaging (MRI) and found that stumps with a high signal had a high retear rate.¹⁶ Shinohara et al also observed that these stumps had extensive inflammatory infiltrates.²⁸ However, the mechanism of retear has been unclear in cuffs that are in poor condition. Furthermore, there have been no reports on postoperative changes in the signal intensity of the rotator cuff tendon stump. Even after surgery with poor-quality tendon stumps, there are few ways to predict retear in the postoperative course. In recent years, predictions of retear with artificial intelligence have been reported^22,27; however, these predictions are only based on preoperative evaluation. If the signs could be observed in the early postoperative period, it may be possible to prevent retear by modifying postoperative rehabilitation.^20,26

In this study, we investigated patients who underwent ARCR using the SB method with the aim of identifying risk factors for retear, including the quality or signal of the tendon stump. We also investigated whether MRI findings 3 weeks after surgery could predict retear cases.

Materials and methods

This study was approved by our institutional review board. All patients provided informed consent. The study had a retrospective cohort design and included patients who underwent ARCR for a tear judged to be medium according to the DeOrio and Cofield classification.^10,12 All shoulders that underwent ARCR after preoperative MRI between April 2016 and March 2023 were enrolled, and shoulders with less than 1 year of postoperative follow-up, with the luck of evaluations, and those with postoperative fall-down were excluded. All surgeries were performed by the single surgeon (S.F., coauthor), who had over 30 years of experience in shoulder surgery. The surgery was performed because of intractable pain and functional disability as a result of rotator cuff tear.

Surgical technique

All surgeries are performed in the beach chair position under general anesthesia. Arthroscopic repair is performed using the SB technique,²⁴ with medial knot-tying after SB lateral row repair to avoid the concentration of stress described in the report by Takeuchi et al.³² An abduction brace is used during the first 3 weeks after surgery. Passive range of motion exercises are started on the day after surgery. Active range of motion exercises are started from 9 weeks postoperatively, and resistance exercise is started from 13 weeks postoperatively.

Evaluation and analysis

For each case, age, sex, body mass index, comorbidities (diabetes, dyslipidemia), smoking history, and traumatic injury were investigated as potential risk factors. In all cases, we first investigated the characteristics of signal intensity on a 1.5-T MRI system (GE Healthcare Japan, Hino, Japan). On preoperative MRI, the anteroposterior tear width was measured in the sagittal view and the medial-lateral width in the coronal view. The occupation ratio of the supraspinatus muscle was calculated as the ratio of the area of the supraspinatus fossa in the oblique sagittal Y view measured on MRI to the cross-sectional area of the supraspinatus.¹¹ Fatty degeneration of the supraspinatus was classified using the 5-point grading system devised by Goutallier et al.² The condition of the tendon stump was classified and evaluated based on the rotator cuff/deltoid muscle (C/D) ratio devised by Ishitani et al.¹⁶ The signal intensity is measured within a 10-mm² circle located 5-8 mm (0.19-31 in) from the edge of the stump as the medial anchor suture and at the point of deltoid muscle where a 45-degree line passes through the end of the tendon. Comparing the signal intensity of the deltoid (D) and rotator cuff tear (C), a C/D ratio of <0.8 was defined as type 1, 0.8-1.3 as type 2, and >1.3 as type 3.

MRI evaluations were performed at 3 weeks postoperatively (at the time of brace removal), at 6 months, and 1 year in all cases. The C/D ratio was evaluated at the end of the sutured cuff in the manner used to assess the tendon stump preoperatively. The signal intensity of the tendon was evaluated at the point 5 mm (0.19 in) medial to the end of the tendon (red arrow in Fig. 1). The signal intensity of the deltoid muscle was evaluated at the point where the 45-degree line (yellow line in Fig. 1) passed through the end of the tendon (blue arrow in Fig. 1). The quality of repair was classified into types I to V according to the Sugaya classification system.³⁰ In this study, type I-III shoulders were classified as the repair group and type IV and V shoulders as the retear group. In the cases with retear, the type of retear was evaluated using the classification system developed by Cho et al.⁶ Clinical outcomes were evaluated using the Japanese Orthopaedic Association and the University of California, Los Angeles scores preoperatively and 3, 6, and 12 months after surgery.

Figure 1.

Method used to calculate the signal intensity ratio. The signal intensity of the tendon was evaluated on a coronal T2-weighted fat-suppressed image at the point 5 mm (0.19 in) medial to the end of the tendon (red arrow). The signal intensity of the deltoid muscle was evaluated at the point where a 45-degree line (yellow line) passed through the end of the tendon (blue arrow).

These variables were compared between the repair and retear groups, and risk factors for retear were investigated by univariable and multivariable analyses. The clinical results and signal changes in the sutured cuff were also compared between the two groups. The same comparisons were also performed among the groups with type 1, 2, and 3 tendon stumps.

Statistical analysis

Quantitative data were analyzed using the Mann–Whitney U test or Kruskal–Wallis test as appropriate and categorical data using Fisher's exact test. The repair and retear groups were analyzed by stepwise logistic regression to identify risk factors for retear. All statistical analyses were performed using SPSS, version 28.0 (IBM Corp., Armonk, NY, USA). A P value < .05 was considered statistically significant.

The intraclass correlation coefficient (ICC)²⁹ was used to assess the intra-rater and inter-rater reliability of C/D ratio measurements in 30 randomly selected cases blinded to patient information. One investigator (J.K.) examined the preoperative MRI scans on two occasions separated by an interval of 2 weeks to confirm the intraobserver reliability of the measurements. Another investigator (coauthor, K.Y.) performed the same measurements so that interobserver reliability could be assessed. Reliability was considered excellent when the ICC was >0.9, good when it was 0.75-0.9, moderately reliable when it was 0.5-0.75, and poor when it was <0.5.

Results

A total of 201 shoulders underwent ARCR at our institution during the study period and had preoperative MRI data available. Seventeen shoulders with follow-up for less than 1 year or the luck of evaluations and 1 with postoperative slip-down were excluded, leaving data for 183 shoulders in 175 patients (men, 102 shoulders; women, 81 shoulders). The mean patient age was 67 (range, 43-85) years.

The ICCs confirmed that intra-rater reliability of C/D ratio measurement was moderate (r = 0.55; 95% confidence interval [CI]: 0.25-0.76, P < .001), as was inter-rater reliability (r = 0.71; 95% CI 0.47-0.85; P < .001).

In terms of the structural outcome, 91 cases were classified as Sugaya type I, 64 as type II, 4 as type III, 10 as type IV, and 14 as type V, giving an overall retear rate of 13.1%. In the retear group, two cases were detected on examination at 3 weeks, 21 at 6 months, and 1 at 1 year after surgery. Using the system devised by Cho et al, the retear was classified as type I in 10 cases (41.7%) and type II in 14 (58.3%). The preoperative tendon stump was classified as type 1 in 48 shoulders (26.2%), type 2 in 76 (41.5%), and type 3 in 59 (32.2%).

In univariable analysis, the tear was significantly larger in the retear group than in the repair group in both the anteroposterior direction (mean 19.4 ± 6.3 vs. 14.8 ± 5.9 mm; P = .001) and medial-lateral direction (26.6 ± 7.0 vs. 19.9 ± 7.6 mm; P < .001) (Table I). There was a significant difference in the distribution of the preoperative stump classification between the repair group and the retear group (type 1/2/3 in 47/71/41 vs. 1/5/18, respectively; P < .001). The C/D ratio was significantly higher in the retear group (1.66 ± 0.74 vs. 1.12 ± 0.54; P < .001) (Table I).

Table I.

Results of univariable analysis of potential risk factors between the retear and repair groups.

Variable	Repair (n = 159)	Retear (n = 24)	P value
Preoperatively
Age (yr)	66.7 ± 7.4	68.9 ± 7.9	.59
Sex (male/female)	85/74	17/7	.13
Body mass index	24.3 ± 3.3	26.1 ± 6.0	.30
Trauma-associated tear	55 (34.6%)	5 (20.8%)	.25
Comorbidities
Diabetes	17 (10.7%)	3 (12.5%)	.73
Dyslipidemia	40 (25.2%)	4 (16.7%)	.45
Smoking	58 (36.5%)	8 (33.3%)	.82
Condition of rotator cuff
Occupation ratio	0.51 ± 0.11	0.50 ± 0.10	.83
Goutallier's classification (0/1/2/3/4)	29/114/15/1/0	2/20/2/0/0	.59
Tear width
Anteroposterior	14.8 ± 5.9	19.4 ± 6.3	.001∗
Medial-lateral	19.9 ± 7.6	26.6 ± 7.0	<.001∗
Stump classification (I/II/III)	47/71/41	1/5/18	<.001∗
C/D ratio	1.12 ± 0.54	1.66 ± 0.74	<.001∗
JOA score	70.4 ± 12.1	69.9 ± 12.1	.89
CLA score	18.9 ± 5.0	18.5 ± 5.2	.84
3 weeks postoperatively
C/D ratio	0.84 ± 0.72	1.70 ± 1.47	.002∗
3 mo postoperatively
JOA score	78.4 ± 10.4	80.8 ± 10.5	.22
UCLA score	24.6 ± 6.4	26.0 ± 5.6	.40
6 mo postoperatively
JOA score	88.4 ± 6.2	85.5 ± 9.2	.13
UCLA score	30.6 ± 4.2	29.6 ± 4.4	.17
C/D ratio	2.00 ± 1.22	-	-
1 yr postoperatively
JOA score	93.1 ± 4.5	90.5 ± 7.2	.18
UCLA score	33.1 ± 2.2	31.9 ± 3.5	.10
C/D ratio	1.52 ± 1.19	-	-

C/D, rotator cuff/deltoid muscle; JOA, Japanese Orthopaedic Association; UCLA, University of California, Los Angeles.

Data are expressed as the number (percentage) or as the mean ± standard deviation.

∗P < .05.

Independent risk factors for retear identified in multivariable analysis were the medial-lateral tear width (odds ratio: 0.88; 95% CI: 0.82-0.95; P < .001) and C/D ratio (odds ratio: 0.30; 95% CI: 0.16-0.58; P < .001).

Patient characteristics were comparable between the types of tendon stump, apart from a significant difference in the smoking rate (18.8% for type 1, 40.8% for type 2, and 44.1% for type 3; P = .013) (Table II).

Table II.

Results of univariable analysis of potential risk factors according to stump type.

Variable	Type I (n = 48)	Type II (n = 76)	Type III (n = 59)	P value
Preoperatively
Age (yr)	67.5 ± 6.0	66.0 ± 8.2	67.8 ± 7.5	.49
Sex (male/female)	22/26	43/33	37/22	.21
Body mass index	24.2 ± 3.0	24.3 ± 3.2	25.0 ± 5.0	.86
Trauma-associated tear	16 (33.3%)	23 (30.3%)	21 (35.6%)	.80
Comorbidities
Diabetes	7 (14.6%)	7 (9.2%)	6 (10.2%)	.63
Dyslipidemia	14 (29.2%)	14 (18.4%)	16 (27.1%)	.32
Smoking	9 (18.8%)	31 (40.8%)	26 (44.1%)	.013∗
Occupation ratio	0.51 ± 0.11	0.51 ± 0.12	0.49 ± 0.1	.34
Goutallier's classification (0/1/2/3/4)	10/35/2/1/0	14/52/10/0/0	7/47/5/0/0	.28
Tear width
Anteroposterior	15.0 ± 5.9	14.8 ± 5.3	16.5 ± 7.1	.24
Medial-lateral	20.0 ± 8.1	20.4 ± 7.7	21.9 ± 7.8	.30
C/D ratio	0.60 ± 0.11	1.04 ± 0.13	1.88 ± 0.55	<.001∗
JOA score	70.9 ± 12.5	70.9 ± 12.3	69.2 ± 11.5	.65
UCLA score	19.0 ± 5.3	19.3 ± 5.1	18.3 ± 4.7	.56
3 weeks postoperatively
C/D ratio	0.59 ± 0.40	0.95 ± 0.74	1.22 ± 1.19	<.001∗
3 mo postoperatively
JOA score	77.5 ± 10.4	76.9 ± 11.3	82.4 ± 8.1	.007∗
UCLA score	23.8 ± 6.0	23.8 ± 6.7	27.1 ± 5.4	.012∗
6 mo postoperatively
JOA score	88.5 ± 6.4	87.5 ± 7.4	88.2 ± 6.0	.91
UCLA score	30.4 ± 4.3	30.2 ± 4.8	30.9 ± 3.3	.95
C/D ratio	2.13 ± 1.77	1.82 ± 0.78	2.17 ± 1.00	.15
1 yr postoperatively
JOA score	94.2 ± 3.7	92.1 ± 5.6	92.4 ± 4.9	.11
UCLA score	33.5 ± 1.8	32.8 ± 3.0	32.8 ± 2.2	.09
C/D ratio	1.31 ± 0.72	1.62 ± 1.29	1.58 ± 1.39	.19

C/D, rotator cuff/deltoid muscle; JOA, Japanese Orthopaedic Association; UCLA, University of California, Los Angeles.

Data are expressed as the number (percentage) or as the mean ± standard deviation.

∗P < .05.

After surgery, the clinical scores were significantly different at only the 3-week assessment: 77.5 ± 10.4 for type 1, 76.9 ± 11.3 for type 2, and 82.4 ± 8.1 for type 3 (P = .007) using the Japanese Orthopaedic Association system and 23.8 ± 6.0, 23.8 ± 6.7, and 27.1 ± 5.4, respectively (P = .012) using the University of California, Los Angeles method (Table II). The C/D ratio was significantly higher in the retear group than in the repair group both before surgery (1.66 ± 0.74 vs. 1.12 ± 0.54; P < .001) and at 3 weeks postoperatively (1.70 ± 1.47 vs. 0.84 ± 0.72; P = .002) (Table I, Fig. 2). The mean C/D ratio varied significantly according to the type of tendon before surgery (type 1, 0.60 ± 0.11; type 2, 1.04 ± 0.13; type 3, 1.88 ± 0.55; P < .001) and at 3 weeks postoperatively (0.59 ± 0.40, 0.95 ± 0.74, and 1.22 ± 1.19, respectively; P < .001), but subsequently showed similar values (Table II, Fig. 3).

Figure 2.

Comparison of signal change in the tendon stump between the repair group and the retear group. The C/D ratio was significantly high in the retear group until 3 weeks postoperatively and remained higher in the repair group at 6 months postoperatively. ∗P < .01. C/D, rotator cuff/deltoid muscle.

Figure 3.

Comparison of signal change in the tendon stump between the types of tendon. There was a significant difference among the tendon types until 3 weeks postoperatively that disappeared during the later postoperative course. ∗P < .001. C/D, rotator cuff/deltoid muscle.

Discussion

In the current study, we identified medial-lateral width and the C/D ratio to be independent risk factors for retear after ARCR even in medium size tear.

Ishitani et al¹⁶ developed a system for classifying the tendon stump using the ratio of the signal intensity at the site of the torn tendon to that in the deltoid muscle on coronal T2-weighted fat-suppressed MRI. In their series, 19.3% of cases were type 1, 40.0% were type 2, and 40.7% were type 3. They also found that the retear rate after ARCR was significantly higher for a type 3 stump and identified stump classification as an indicator of rotator cuff fragility. Our present study found a higher prevalence of type 1 stumps, which may reflect the fact that only medium rotator cuff tears were included. Takeuchi et al investigated 389 patients who underwent ARCR and identified a type 3 tendon stump to be an independent predictor of retear with both the SB and double-row suture technique.³¹ In this study, we investigated the risk factors for retear of a repaired rotator cuff, including the tendon stump, and also evaluated the postoperative change in C/D ratio in the tendon. Overall, in our patients, all of whom had a medium tear, the retear rate was 13.1%, which is not high compared with previous reports.^8,9,14,18,24 Choi et al reported a retear rate of 10.6% in a similar study that focused on medium tears,⁹ and Park et al found a retear rate of 13.3% for small to large tears.²⁵ A systematic review by Longo et al reported a retear rate of 12.5% for small to interrupted tears.²¹ The timing of the retear was within 6 months in almost all of our cases, which is consistent with previous reports.^15,23

Our study identified medial-lateral width and the C/D ratio to be independent risk factors for retear after ARCR. A systematic review identified age, sex, body mass index, and comorbidities as risk factors,³⁴ but no significant between-group differences in these factors were found in our study. However, our detection of medial-lateral width as a risk factor is consistent with previous reports.^1,4,5,7 Furthermore, our identification of stump condition as a risk factor is in line with the report by Ishitani et al.¹⁶ Therefore, stump status must be recognized as a risk factor even in patients with a medium tear.

In the study by Shinohara et al, type 3 stumps showed a higher accumulation of advanced glycation end products, increased oxidative stress and apoptosis, and reduced cell viability, resulting in a fragile rotator cuff,²⁸ meaning that a high C/D ratio reflects degeneration and fragility of the rotator cuff. Cho et al observed that when the tendon was of poor quality after use of the SB method, the tendon portion was as weak as the musculotendinous junction, becoming vulnerable to retears at the original repair site as with single-row repair,⁶ which is consistent with our results.

We found a statistically significant difference in the smoking rate among the various tendon types. In another study, histological analysis revealed an association of smoking with increased inflammation and marked accumulation of fat and fibrogenesis in the cuff.¹⁹ Shinohara et al found that expression rates of reactive oxygen species and apoptosis were significantly higher in type 3 stumps than in the other types.²⁸ These reports could provide a rationale for our results. However, there were no differences in the clinical scores among the stump types. In terms of the relationship between the type of stump and clinical outcomes, in a study of 75 patients who underwent arthroscopic superior capsule reconstruction, Ben et al found that patients with a type 1 stump had a significantly higher general score and good forward flexion.² Their findings are in contrast with our present results.

We found that the C/D ratio was higher at 3 weeks postoperatively in the retear group. If a high C/D ratio increased oxidative stress and associated apoptosis, as mentioned earlier, C/D ratio change could be a predictor of retear. Therefore, MRI examination during the early postoperative period could be useful for detecting signs of retear.

However, the C/D ratio showed similar values among the tendon types after the 3-week assessment and became high by 6 months after surgery, even in the repair group. During the proliferative and remodeling stages of tendon healing in particular, tendon fibroblasts synthesize collagens, proteoglycans, and other components of the extracellular matrix.¹⁷ The collagen is organized into mature fibers that are subsequently reorganized during the remodeling phase.¹⁷ A postoperative C/D ratio change in the cuff might enhance the remodeling process necessary for tendon healing in the late phase.

Limitations

First, this study had the limitations inherent in all retrospective studies. However, we analyzed prospectively collected data on surgical procedures performed by a single surgeon. Second, our analysis was limited to evaluation of the signal intensity in the cuff, and the histological behavior of the repaired cuff was unclear. Third, the ICC of the C/D ratio measurement was only moderate. We must improve the measurement method in the clinical situation.

Conclusions

We have identified increased C/D ratio and tear width as risk factors for retear after ARCR using the SB method. A higher C/D ratio was detected in retear cases both postoperatively and preoperatively. Stump status must be recognized as a risk factor even in patients with a medium tear.

Disclaimers

Funding: No funding was disclosed by the authors.

Conflicts of interest: The authors, their immediate families, and any research foundation with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.