Introduction

Traumatic anterior shoulder instability is routinely seen in younger patients and its immediate post-reduction complication is relatively low; however, it is known to re-dislocate in 30–90% of young patients [1–3]. In spite of the isolation of the risk factors and the selection of the patients for nonoperative treatment [4], it persists an important rate of after primary anterior shoulder dislocation.

Malgaigne was the first surgeon to describe the humeral bony lesion in shoulder dislocation in 1855 [5]. However, H.A. Hill and M.D. Sachs [6] in 1940 gave a detailed description of this posterolateral humeral bony.

Various radiographic studies [7–9] described different projections to visualize the Hill–Sachs lesion with an accuracy of 90–100%. In the last one, Ito [10] used a new X-ray to achieve an undistorted radiographic view of the notch. Recently, MRI (Magnetic Resonance Imagery) and CT (Computer Tomography) arthrography has also been used increasingly as a precise and reproducible diagnostic tool for a detailed three-dimensional evaluation of the soft tissue and bony lesion following anterior shoulder instability [11]. Roger [12] demonstrated the superiority of CT arthrography for the detection of Hill–Sachs lesion compared to the X-rays. However, these modalities are routinely implemented in combination with the pre-operative planning of the shoulder stabilization procedure.

It has long been recognized that both open and arthroscopic shoulder stabilization procedures leads to good functional outcome. The reported arthroscopic failure rate with different repair technique varies from 49 to 7% [13, 14] and is much higher as compared to the open surgical procedures [15].

In the aim to reduce this high recurrence rate, some studies [16–22] isolated numerous predisposing factors. All these data of the literature are summarized in Table 1.

Table 1 Risk factors for recurrence of shoulder instability after arthroscopic surgery, data of the literature
Full size table

For many authors [4, 16, 17, 19, 23–26], the Hill–Sachs lesion is an important risk factor and its ignorance or its negligence into pre-operative planning could be a reason for arthroscopic procedures failure.

It persists a semantics problem to considerate this humeral bone defect. H.A. Hill and M.D. Sachs [6] classified as small, medium and large on the basis of its diameter. Burkhart [23] described the Hill–Sachs lesion into engaging and nonengaging lesion. For Rowe [26] and Boileau [17], a higher post-surgical failure rate associated with ‘large’ Hill–Sachs lesion, and for Balg [16], one of the criteria to calculate the instability severity index score is an Hill–Sachs lesion ‘visible on external rotation’ on anteroposterior X-ray view. All these definition are subjective and rough.

Kaar [27], in a cadaveric study, found with increasing size of Hill–Sachs defects glenohumeral stability decreasing.

These conclusions raised the problem of quantification and location of Hill–Sachs lesion. In spite of some dedicated articles [4, 24, 25], there is no standard reference.

Hence, the purpose of this study was to assess the influence of the Hill–Sachs lesion on the long-term success of the arthroscopically stabilized unstable shoulder. We hypothesized that the importance of the Hill–Sachs lesion has a predictive value for the result of arthroscopic stabilization procedure and could be used as a selection criteria along with glenoid lesion to choose between arthroscopic and open technique. We, finally, looked for a simple and reproducible radiographic measure to quantify the Hill–Sachs lesion and a value threshold by making a precise risk factor for failure after an arthroscopic stabilization procedure.

Materials and methods

Seventy-one consecutive patients (49 men and 22 women), with traumatic anterior shoulder instability who had arthroscopic stabilization procedure at our institution, were included in this study. The mean age of the patient at the time of arthroscopic procedure was 29 years, and the dominant shoulder was involved in 27% of the cases. The mean follow-up was 68 months (32–100 months).

Forty-two patients (59%) were involved in sports, with twenty-two (31%) participated in high risk sports with or without cocked position of the arm, thirteen (18%) involved in overhead sports and the remaining seven (10%) participated at competitive level.

The patients were assigned to three groups on the basis of their symptoms and physical examination finding: Recurrent dislocation group, patients with 2 or more episodes of dislocation requiring close reduction (41 patients); recurrent subluxation group, patients with recurrent subluxation with spontaneous reduction (21 patients) and painful unstable shoulder group, patients with history of pain and instability following trauma but without dislocation or subluxation (9 patients).

All the patients were subjected to standard anteroposterior radiographs in neutral rotation, 45° external rotation and 45° internal rotation as well as a comparative Bernageau view pre-operatively. Next transparent circular gauges in 1-mm increment were fitted on the humeral head on the 45° internal rotation anteroposterior radiograph, and a best fit circle was used to measure the radius (R) of the humeral head and the depth (D) of the lesion (D) (Fig. 1). Because of the variation in the magnification and to reflect the true size of the notch on the X-ray, a ratio between the depth of the lesion and the radius of the humeral head (D/R) was determined.

Fig. 1
figure 1

Radiographs of an anteroposterior view of the shoulder with measures of the head Radius (R) by means of a transparent gauge and measures of notch Depth (D)

Full size image

A double-contrast computerized arthrotomography of the affected shoulder joint was also carried out for accurate evaluation of the Hill–Sachs lesion. The computed tomography scan was made with a 3-mm slice thickness after injection of 1 mL (millilitre) of contrast medium and 19 mL of room air into the glenohumeral joint.

The CT scans were evaluated by an independent observer to measure the following parameters on the humeral head (Fig. 2).

Fig. 2
figure 2

CT scan axial slide with measures of the Head Diameter (HD), Width and Depth (De) of the Hill–Sachs lesion

Full size image
  1. 1.

    Humeral head diameter (HD): represents the maximum anteroposterior distance, determined by a best fit circle along the circumference of the humeral head which also gives us the centre of the circle.

  2. 2.

    Width of the lesion (W): represents the distance between the medial and lateral edge of the lesion, determined by two trajectories projected from the centre of the humeral head along the edges of the lesion.

  3. 3.

    Depth of the lesion (De): represents the distance between the apex and bottom of the lesion along the best fit circle.

  4. 4.

    Length of the Hill–Sachs lesion (L): represents the distance between the initial and the last CT slice where the lesion is visualized.

  5. 5.

    Lateralization angle: measured for precise location of the lesion and determined by a best fit circle on the humeral head and by an intersection of an anteroposterior axis through the centre of the humeral head and another extending from the centre of the circle to the centre of the lesion (Fig. 3). This method of measurement was described by Richards et al. [28] on shoulder MRI.

    Fig. 3
    figure 3

    CT scan axial slide with measure of the angulation of the Hill–Sachs lesion

    Full size image

Assuming the Hill–Sachs lesion as an irregular geometrical body and for the purpose of calculation, we applied the formula of an ellipsoid for calculating the Hill–Sachs lesion volume.

The arthroscopic stabilization procedure was carried out as a day case in a lateral decubitus position under general anaesthesia and interscalene block. Most of the patients had transglenoïd fixation, and 23% of patients also had a capsular shift with Bankart repair. Post-operative recovery was uneventful. The patients were managed with the arm in a sling in internal rotation for 4 weeks. Passive pendulum exercises were started on the day after surgery. Shoulder rehabilitation with a physiotherapist was commenced after a month. External rotation was limited to 45º until 45 days. Re-creational and over-head sports were allowed after 3 and 4 months, respectively.

The post-operative functional results were evaluated by an independent observer with the Duplay group clinical score.

Statistical analysis

To evaluate the reliability of X-ray measurements, 10 orthopaedic surgeons independently measured the depth of the lesion and the radius of the humeral head in 10 patients on the anteroposterior X-ray in 45° internal rotation. Inter-Observer reliability of the measurement was tested by calculation of the intra-class correlation coefficient (ICC).

We used the student’s t test, Tukey’s post-test, ANOVA (ANalysis Of Variance), Chi-square test and Mann–Whitney test. The level of significance was set at P < 0.05. Data were analysed with Statview 5.0 (SAS institute, Cary, North Carolina).

Results

Fifty-nine patients were available for final follow-up review. The mean follow-up was 68 months (32–100 months). The functional results were classified as excellent (24 patients), good (14 patients), fair (7 patients) and poor (14 patients). For statistical purpose, the first two and the last two functional groups were combined. Nine patients from the poor group had previous open Bankart procedure, 7 for recurrence of dislocation and 2 for recurrence of subluxation. There was no significant difference in patient demographics between the combined excellent and good group (38/59) and the fair and poor group (21/59).

The Hill–Sachs lesion was seen as a notch in fifty patients (84%) on the anteroposterior radiograph with the arm in 45° internal rotation (Fig. 1), whereas on Bernageau view, glenoïd pathology was seen in forty patients (68%). Glenoïd surface abrasion was seen in 29 patients (49.15%), nine patients (15.25%) had an antero-inferior glenoïd fracture with a detached bone fragment and two patients (3.38%) had amputation (defect) of the anterior glenoïd surface. The following Table 2 shows different glenoïd pathology in the two functional groups. This glenoïd lesions had no statistically significant influence on the overall surgical result except in three patients (5%) with recurrence of instability, a significant antero-inferior glenoïd lesion (two amputations and one fracture) was present but on the humeral side a small Hill–Sachs lesion with a volume was less than 500 mm³ was present.

Table 2 Results of glenoid lesion from Bernageau view in the two functional groups
Full size table

The mean D/R ratio was 16.2% (3–34%). Tukey’s post-test showed a significantly high D/R ratio in recurrent dislocation versus the subluxation group (Table 3). This ratio was also statistically significant in the two functional groups (Table 4). The inter-observer correlation coefficient values for the depth of the lesion and the radius of the humeral head were 0.67 and 1.31 mm, respectively, which was statistically insignificant.

Table 3 Results of Hill–Sachs lesion parameters in pre-operative symptomatic group as measured from X-ray and computerized arthrotomogram
Full size table
Table 4 Results of Hill–Sachs lesion parameters from X-ray and CT arthrogram in the two functional groups
Full size table

The mean Hill–Sachs lesion volume was 1019 mm³ (0–4,792 mm³). This lesion volume represented 2.28% of the head volume (range 0–10.5%). The lesion volume variation in the 3 patient groups was statistically significant (ANOVA P = 0.001) (Table 3). This volume was also significantly high in the fair and poor functional group (Table 4). The mean lesion/head volume ratio (VR) in the two functional groups was 1.5% (0–8.4%) and 4.9% (0.6–10.5%), respectively.

The lateralization angle measured to precisely locate the lesion ranged from 165° to 218° with a mean of 188° in the excellent and good group and 193° in the fair and poor group that was statistically insignificant. However, this angle was significantly different in the three symptomatic groups with much higher angle in the recurrent dislocation group.

The instability recurred in ten patients (16.9%) following arthroscopic stabilization procedure, which were later revised with open coracoid transplantation. Majority of the recurrences were from the recurrent dislocation group with significantly high D/R and volume of the Hill–Sachs lesion.

Discussion

This study found that arthroscopic stabilization procedure in patients with high volume of the Hill–Sachs lesion led to a post-operative recurrence of instability in 16.9% of patients with a mean follow-up of 68 months. This result is comparable to those found in the other similar studies [18]. Few authors reported lower failure rate but they had a less important average follow-up [16, 17, 21, 22], and or the radiographic lesions (glenoïd and Hill–Sachs) were criteria of exclusion [22]. The long mean follow-up of our series was essential because it was recently clearly proved [19] the importance of this risk factor in the analysis of the results of this surgery.

The most important finding from this study was a simple measure on conventional X-rays and reproducible in the daily clinical practice: D/R. When the D/R ratio threshold was more than 15%, the failure rate was 56% contrary to only 16% failure when the D/R ratio was less than 15%. Patients who had a high D/R ratio as seen in the recurrent dislocation group had significantly higher rate of recurrence of instability.

At the same time, the Hill–Sachs volume was significantly high in patients with history of recurrent dislocation leading to 37% recurrence in this group. When the lesion volume was compared in the two post-operative functional groups, we found it to be significantly high in the group which performed poorly. When the lesion volume threshold was considered to be 1,000 mm³, the surgical failure rate below this threshold was 7.7% and dropped to 2.5% on exclusion of the cases with glenoïd fractures and amputations.

Even if recently Kaar [27] proved glenohumeral joint stability decreases with increasingly larger humeral head defects and Cetik [24] correlated the number of dislocations and the extent and depth of the Hill–Sachs lesions; nobody of them gave morphometric details of the Hill–Sachs lesion.

Ito [25] first, on X-ray described by himself [10], and Kralinger [4], on anteroposterior with 60° of internal rotation, tried to quantify the humeral bon defect. In their analysis, they did not take into account the radiographic magnification, which always exists and did not show any relationship between the Hill–Sachs lesion and the incidence of recurrence after arthroscopic stabilization procedure.

As Saito [29], we thought that not only the size but also the location of the Hill–Sachs lesion determined the risk factor for recurrence. This angle was significantly much higher in the recurrent dislocation group but no statistically significant difference could be found between combined excellent/good group and combined fair/poor group. It can be explained by the fact that only one computerized arthrotomography slice showing the deepest position of the notch was taken into account while measuring this angle, possibly giving a false position of the notch. However, if we had measured this angle on all the slices showing the lesion would have given the obliqueness of the notch and may have given a different lateralization angle.

Several authors [11, 12] found 3D (Dimension) CT reconstruction of the glenohumeral joint useful to visualize the Hill–Sachs lesion and calculate its volume, whereas Hammar [30] used ultrasound scan for measuring the volume. Saito [29] thinks the CT scan may overestimate the Hill–Sachs lesion because it is sometimes difficult to distinguish from the bare area.

Various methods like capsuloplasty, mosaïcplasty and proximal humeral osteotomy have been advocated to reduce the incidence of surgical failure secondary to Hill–Sachs lesion. However, capsuloplasty is not an acceptable procedure in a young population with high functional demand as it leads to limitation in the range of glenohumeral external rotation. Mosaïcplasty has limited indication and is reserved for large defects. Proximal humeral rotation osteotomy described by Weber leads to subscapularis and capsular shortening and is also associated with complications like pseudarthrosis, implant failure and axillary nerve injury.

Hence, stabilization procedures like Bankart repair open or arthroscopic are the preferred methods to deal with anterior shoulder instability. Although the incidence of failure is high with arthroscopic stabilization procedure, it is still developing and hopefully in future the recurrence of instability will be much lower.

Conclusion

The results from this study suggest that pre-operative radiological assessment of the Hill–Sachs lesion can predict success of arthroscopic surgical repair. Arthroscopic stabilization procedure should be reserved for Hill–Sachs lesion with a D/P less than 16% but any associated glenoïd lesions should also be quantified.

High volume Hill–Sachs lesion and glenoïd bony defects clearly compromise the success of arthroscopic stabilization procedure, and hence, the management of anterior shoulder instability must include proper patient selection.