Introduction

In 2004, Wolf and Pollack [22] described the arthroscopic technique of Hill-Sachs remplissage associated with a Bankart repair (BHSR) and ushered the perspective of a new option to treat anterior shoulder instability with the capsulotenodesis of the infraspinatus into the humeral head defect as an augmentation of the Bankart repair [23].

Most authors considered that the main criteria to indicate a Bankart repair reinforced with a remplissage procedure was the size of the Hill-Sachs lesion [2, 22, 23]. More recently the “glenoid track” concept took into account combined criteria as the position, the width and glenoid bone lost associated with the Hill-Sachs lesion. It depicted differently Hill-Sachs lesions that should be addressed because they could engage onto the glenoid rim and be responsible of a failure of a Bankart repair alone [9, 24]. The literature reported short-term promising outcomes after this surgical technique with a failure rate of less than 10% and no major complication [2, 4, 6, 10, 14, 15, 17, 25].

We hypothesised that arthroscopic BHSR performed primarily was a reliable option to treat chronic anterior shoulder instability without substantial glenoid bone loss. Thereby, the aim of this study was, firstly, to evaluate clinical outcomes at mid-term follow-up of this procedure and, secondly, to identify prognostic factors of failure.

Materials and methods

Patient selection

In a prospective design study, we included patients who had surgery between 2009 to 2013. Inclusion criteria were: (1) patient operated on for traumatic anterior shoulder instability with an Instability Severity Index Score (ISIS) above 3 points [1], (2) no substantial glenoid bone loss (<25%), (3) treated with an arthroscopic Bankart repair associated with a Hill-Sachs remplissage and (4) followed for a minimum of 24 months. Exclusion criteria were: (1) revision surgery, (2) voluntary instability in any direction and (3) first-time dislocation or subluxation. Patients gave their written agreement prior the enrolment and this study was approved by our local institutional review board.

Surgical technique

The procedure was performed by a single surgeon with the patient in a beach-chair position under general anaesthesia with an interscalene nerve block. Three different portals were used for the arthroscopic BHSR procedure: a standard posterior portal and anterosuperior portals for the Bankart repair and capsularshift, and an accessory posterolateral out-in portal for the remplissage of the humeral bone defect.

Firstly, with a 30° scope introduced in the posterior portal, the joint was explored, confirming Bankart lesion, the amount of glenoid and/or humeral bone defect, and antero-inferior capsular redundancy. Dynamically with abduction and external rotation of the arm, the “engaging” nature of Hill-Sachs lesion was explored [5]. Then, the labrum was mobilised from the two to six o’clock position and the anterior glenoid rim was debrided in order to prepare the Bankart repair. The anterior capsule was shifted upward with a temporary suture traction as described by Boileau et al. [3]. The first single loaded, impacted and absorbable anchor (Gryphon BR; DePuy Mitek, Raynham, MA, USA) was placed at the five o’clock position without tightening the suture in order to facilitated the next step.

Secondly, the scope was switched from posterior to anterior portal and the accessory posterolateral portal was created with a spine needle oriented perpendicular to the Hill-Sachs lesion. A posterior to anterior manual translation of the humeral head enhanced the view and the access to the posterior chamber. A working 6.5-mm cannula was pushed through the deltoid muscle and was positioned outside the joint, in the sub-deltoid space. The Hill-Sachs lesion was abraded directly with the drill guide used to insert the anchors. Two single loaded, impacted and absorbable anchors (Gryphon BR; DePuy Mitek) were placed in the valley of the defect, one superior and one inferior, through transtendinous passage. Then, with a penetrating grasper, one limb of each suture was retrieved through the posterior aspect of the capsule and the infraspinatus tendon, in order to prepare two mattress stitches (tied at the last step of the procedure).

Thirdly, after switching back the scope in the posterior portal, the Bankart repair and the capsular shift was performed with a minimum of two others anchors placed at the four and three o’clock positions [3]. Then, the Hill-Sachs lesion was filled by tying blindly the two last mattress stitches one the bursal side of the infraspinatus, creating a capsulotenodesis (Fig. 1a-c).

Fig. 1
figure 1

Capsulotenodesis (scope placed in anterior portal). a Before remplissage. b After placement of anchors and sutures retrieved through the posterior cannula. c Result of remplissage after mattress stitches achieved

Full size image

Post-operative care

The arm was placed in a sling in internal rotation for six weeks, with no external rotation above the neutral position during this period. Self pendulum exercises were required immediately after the surgery and formal physiotherapy was delayed at three weeks. Sports activities without risk for the shoulder were allowed at three months, and overhead or contact sports at six months post-operatively.

Clinical evaluation

The patients were prospectively evaluated at six weeks, three months, six months and every year. After two years’ follow-up, clinical assessment was done during a specific visit by an independent observer. Active ranges of motion were measured pre-operatively and post-operatively, in forward flexion, external rotation elbow at side and at 90° of abduction, and internal rotation (level of the thumb on the spine). Pre-operative hyperlaxity was defined as an external rotation elbow at the side greater than 85° and/or a positive Gagey test [1]. An apprehension test in abduction external rotation was performed post-operatively at each review, beyond the third month. Objective functional results were evaluated with Rowe and Walch-Duplay scores, including sports activity, motion, pain, functional results, and stability [18, 21]. Subjective results were based on Simple Shoulder Value (SSV), ranging from 0 to 100% [12].

Radiographic evaluation

Pre-operatively, X-rays (anteroposterior and Bernageau views) and computed tomography (CT) scan (with three-dimensional reconstruction) or arthro-CT scan were systematically required to assess bone defects. On CT scan, Hill-Sachs lesion was qualified pre-operatively as “on-track” or “off-track” [9]. Its depth was quantified as the ration with the humeral head diameter. Glenoid bone defect was analysed using the Sugaya’s method [20].

Post-operatively, X-rays were performed at each review and allowed to follow radiographic evolution of absorbable anchors used for the Bankart repair and the Hill-Sachs remplissage. Signs of osteolysis around the anchors were carefully looking at. Glenohumeral osteoarthritis was classified according to Samilson and Prieto classification [19].

Statistical analysis

Measurements were expressed as the mean, range and standard deviation. The D’Agostino-Pearson test was used to analyse data distribution. Paired values were compared by using the paired t-test, and unpaired results were compared by using the Mann-Whitney test. The chi-squared or Fisher’s exact test was used to compare categorical data. The significance level was set at p < 0.05.

Results

Patient population

From November 2009 to January 2014, 191 patients were operated on for traumatic anterior instability of the shoulder in our department. Forty-three patients were treated with an arthroscopic BHSR. Thirty-seven patients fulfilled the inclusion criteria and three patients were excluded because of revision surgery, leaving 34 patients (34 shoulders) available for this study.

The mean age at the time of surgery was 26 years old (range, 15-49; ± 8.5); there were 30 males and 4 females. Twenty-four (70%) patients participated in at least one sports activity and 7 (30%) at a competitive level: 1 patient in a sports category without risk for the shoulder, 14 patients in a contact-sports category and nine patients in an overhead-sports category. There were signs of hyperlaxity in 13 (38%) shoulders and the mean preoperative ISIS was 4.2 ± 1.9 points (range, 3-9).

The dominant shoulder was involved in 13 cases (38%) and the mean number of dislocation and/or subluxation before the procedure was respectively 3.4 ± 4 (range, 0-20) and 11.7 ± 24 (range, 0-100) respectively. Only two patients did not report frank shoulder dislocation requiring a reduction manoeuvre in the emergency department and presented subluxation of their shoulder. The mean time between the first episode of instability and the procedure was 59 ± 67 months (range, 6-240).

On pre-operative radiographic analysis, all Hill-Sachs lesion were “on-track” and the mean depth was 20.1% ± 8.7 (range, 7-40). Only 11 (32%) patients had a bone loss on the glenoid side, with a mean of 12.1% ± 5.9 of the glenoid width (range, 5.4-23.2).

Overall functional results

At a mean follow-up of 35 months (range, 24–63), the mean Rowe and Walch-Duplay score achieved 92.7 ± 10.5 points (range, 58-100) and 88.2 ± 18.4 points (range, 35-100) respectively.

Comparing active range of motion pre-operatively and at last follow-up, the deficit in external rotation elbow at side and at 90° of abduction was of 6° and 1° respectively (p > 0.05) (Table 1).

Table 1 Ranges of motion
Full size table

Return to sport

Post-operatively, 18 (53%) patients resumed the same sports activities and 15 (44%) exactly at the same level. The delay to return to sport was 6.5 ± 3.4 months (5–13).

Recurrences

Five (14.7%) patients had a recurrence of instability with at least one subluxation or dislocation, and three others had persistent apprehension or pain in abduction external rotation during clinical examination without true instability. The five failures occurred during the firstyear period following the procedure (mean 10 months; range, 6-12). In the postoperative period, these patients recovered significantly faster range of motion than patients who remained stable overtime especially at 1.5 months (Fig. 2a-d). No significant correlations were found with the amount of glenoid bone loss, hyperlaxity, pre-operative ISIS, age at surgery, gender, sport activity and the risk of recurrent instability. However, unstable shoulders had a deeper Hill-Sachs lesion compared with stable shoulders (26% vs 19%) (Table 2).

Fig. 2
figure 2

Comparison of active mobilities in the stable and unstable groups of patients during post-operative follow-up period. a Active anterior elevation (°). b External rotation at side (°). c External rotation in 90° of abduction (°).d Internal rotation (points)

Full size image
Table 2 Analysis of risk factors of recurrence at follow-up
Full size table

Complications and reoperation

No iatrogenic post-operative complications were reported. One patient described a persistent posterior shoulder pain in abduction beyond six months after the surgery. An MRI identified a bursitis and inflammatory aspect of the infraspinatus muscle-tendon junction.

Three-fifths of the unstable patients after the primary surgery required a revision with an open Latarjet procedure without complication. All of them were stable with no apprehension at follow-up.

Radiographic analysis

At follow-up, two patients had signs of osteoarthritis and were classified type 2 according Samilson and Prieto classification (6%). Osteolysis around the anchors used for the remplissage was identified in two cases on Bernageau view, without clinical consequences.

Discussion

The purpose of this study was to evaluate clinical outcomes of arthroscopic BHSR in case of anterior shoulder instability. At mean follow-up of 35 months, the rate of recurrence reached 14.7% and two main predictive factors of failure were identified: the first was related to the deepness of the Hill-Sachs lesion and the second was related to the way of post-operative range-of-motion recovery.

The rate of recurrence after BHSR commonly reported was less than 10% [2, 4, 6, 14, 15, 17, 25]. Wolf et al. [23] found only 4% of recurrence at two to tenyear follow-up, even in case of critical glenoid bone defect. More recently, Garcia et al. [11] reported a rate of 12% at an average of five year follow-up. Revision surgery and increased preoperative number of dislocations were identified as predictive factors of failure. With 15% of subluxation at 30 months of follow-up, Park et al. [16] reported the highest rate, in a selected population with Hill-Sachs lesion of more than 25% of the humeral head diameter. Almost at the same follow-up and with a mean depth of 20%, our study supported this findings and gave a value to investigations of Hill-Sachs’ features.

Previously Burkhart et al. [5] described the so-called “engaging” lesion, when it engaged onto the glenoid rim during examination under arthroscopy. This dynamic concept was more accurately explained with the “glenoid track” theory, which took into account the position and the width of the Hill-Sachs lesion related to the amount of glenoid bone loss associated [9, 24]. An “off-track lesion” is at risk to engage and, thereby, should be considered as a predictive factor of recurrence after a Bankart repair alone. In our study, every Hill-Sachs were surprisingly “on-track” and should not be absolutely addressed according this concept. However, our indication of a BHSR was mostly based on pre-operative ISIS (above 3 points), the engaging nature during arthroscopic exploration of the Hill-Sachs lesion and the absence of substantial bone loss on the glenoid side. Respecting these criteria, Boileau et al. [2] reported a failure rate of 2% at short-term follow-up.

The depth of the Hill-Sachs lesion was the only pre-operative predictive parameter of failure of this series. One explanation would be related to non-healing of the capsulotenodesis into the Hill-Sachs lesion, leaving the Bankart repair alone to stabilise the shoulder [7, 8, 15, 16]. According a biomechanical model of the capsulotenodesis fixation into a Hill-Sachs lesion, we have shown that a deep and narrow defect made the fixation more critical (see Appendix). Basically, pull-out forces applied on the tendon are concentrated into the point of fixation of the tendon into the bone when the defect is deep and narrow. As a consequence, tissue healing into the Hill-Sachs lesion would be altered in these conditions.

The second predictive factor was clinical, with a faster range of motion recovery post-operatively in the unstable group of patient. Once again, this would explain a subsequent failure of the capsulotenodesis in the deep humeral defect. Previous studies reported that healing of the capsulotenodesis was successful in up to 75% after a remplissage [2, 14]. Unfortunately, we did not investigate the healing in our study, and could not validate the correlation with success or failure of the procedure.

According to these findings, the indication of BHSR is debatable in case of deep and engaging Hill-Sachs lesion. Degen et al. [8] proved that either Latarjet and BHSR procedures reduced frequency of dislocation in a 25% of humeral defect model on cadavers. However, Latarjet seemed to stabilise more specimens than BHSR (94% vs 84% of testing scenarios), and thereby could be proposed as an alternative procedure. The other option would be to consider a bone graft reconstruction [13].

This study has some limitations. The mean follow-up of 35 months remained too short to clearly determine the rate of recurrence after a BHSR, so that the number of failures would increase after the fifth year. Moreover, the amount of patients included was limited and other predictive factors of failure would probably be highlighted in a larger cohort. We did not perform a power analysis to increase the validity of our statistical methods. The second limitation was related to the lack of investigation in the healing of the capsulotenodesis of infraspinatus. Therefore, we could not correlate failures with this parameter.

Conclusions

The BHSR is a safe procedure to address anterior shoulder instability. However, in this series, the rate of failure at mid-term follow-up is higher than commonly reported. The depth of the Hill-Sachs of more than 20% of the humeral head would be a limit to indicate a BHSR, and an alternative surgery should be considered in this preoperative situation. Moreover, the premature recovery of range of motion would be a clinical indicator of failure at follow-up.