Introduction

The Latarjet procedure has yielded good clinical results in cases of anterior instability with bone loss or recurrent instability [3, 20, 27, 37]. Stable initial fixation of the coracoid to the glenoid is essential to minimise the risk of non-union and to facilitate rehabilitation without concern for construct failure. However, overall complication rates of the Latarjet procedure have been reported to be as high as 30% [14]. Reported complications include recurrent instability, non-union, osteolysis and bone block migration, implant-associated complications and neurovascular issues [18, 32, 38]. Neurologic injuries to the brachial plexus and the musculocutaneous and suprascapular nerves have also been reported [11, 14, 28]. Fixation of the coracoid process is usually achieved with two parallel screws. However, the proximity of the suprascapular nerve to the exit sites of the screws is well documented. Iatrogenic suprascapular nerve injury can be due either to bicortical drilling or to the prominence of the screws [22, 31]. Injury to the suprascapular nerve can also result from malpositioning of the graft, leading to unintended screw paths. However, inadequate screw length may increase the risk of non-union and bone block migration if too short or be the cause of a suprascapular nerve lesion or soft tissue discomfort if too long [6].

Coracoid process non-union is found in up to 20%, depending on the series [7,8,9]. Suprascapular nerve injuries associated to shoulder instability surgery have been reported in up to 6% of cases [26].

The results of bicortical fixation of the coracoid to the scapular neck have been reported in the literature but its superiority over monocortical fixation has not been established. Monocortical fixation involves drilling through the coracoid process and into the anterior cortex of the glenoid neck up to, but not perforating, the posterior cortex of the glenoid neck. If monocortical fixation provides comparable or equivalent stability, the risk of injury to the suprascapular nerve would be reduced substantially.

The aim of this biomechanical study was to compare the load to failure and the mode of failure of coracoid process graft fixation with two monocortical screws vs. two bicortical screws. The primary outcome studied was the force in Newton (N) required to detach the coracoid process from the anterior glenoid rim using perpendicular loads. It was hypothesized that the load to failure of a monocortical fixation technique would be less than that of bicortical fixation technique.

Materials and methods

Fourteen scapula sawbones were obtained (Sawbones®, Composite Scapula, 4th generation). Each sawbone underwent the coracoid process transfer according to the technique published by Latarjet [23]. The coracoid process was sectioned near its base and centred below the equator of the anterior glenoid at the 3–5 o’clock position, 1–2 mm medial to the articular surface. The minimal length of the coracoid bone block was 21 mm. In our study, two 3.5 mm partially threaded metal screws were used for the fixation of the coracoid process to the glenoid neck as per Patte et al. [30] and more recent studies [1, 36].

Seven scapula sawbones received a monocortical fixation of the coracoid process to the glenoid, with two parallel 3.5 mm partially threaded solid metal screws measuring 36 mm in length (Synthes, Solothurn, Switzerland, Fig. 1a). A further seven scapula sawbones received a bicortical fixation of the coracoid block with two parallel 3.5 mm partially threaded metal screws measuring 40 mm in length (Synthes, Solothurn, Switzerland, Fig. 1b). The mono- and bicortical length of the screws was confirmed by direct visualisation.

Fig. 1
figure 1

a Monocortical fixation of the coracoid process. b Bicortical fixation of the coracoid process (after fracture of the glenoid, clamp still attached)

Full size image

Biomechanical testing

The scapula Sawbone was solidly fastened in a modified anchor, locking the shoulder blade parallel to the ground. A modified metal clamp was attached to the coracoid block. The traction cycles to precondition the construct and the increasing tensile load were chosen after the work published by Weppe at al [36]. The traction force was applied perpendicular to the coracoid process graft and the force was measured by use of a universal materials testing machine (MTS, MiniBionix 858 II, Eden Prairie, MN, USA, Fig. 2). The testing machine was equipped with a 2.5 KN force sensor with a maximum force error of 0.2%. One hundred traction cycles between 0 and 20 N were performed to precondition the construct. After preconditioning, a break of 30 min was carried out. The coracoid process fixation was then tested to failure by application of an increasing tensile load at a rate of 10 N/s. Failure was defined as a pullout of the screws, detachment of the coracoid process from the glenoid, or fracture of the glenoid. The ultimate failure load and mode of failure were documented. The primary outcome studied was the force in Newton (N) required to detach the coracoid process from the anterior glenoid rim. Approval for the study was obtained from our institution ethics committee. IRB approval: 1333–2012.

Fig. 2
figure 2

Experimental set-up (left scapula)

Full size image

Statistical analysis

The Mann–Whitney U test was used to compare the failure loads of the two fixation techniques (continuous, non-parametric variable). The Fisher’s exact test was used to compare the mode of failure of the two groups (categorical variable). A p value < 0.05 (two tailed) was considered to be statistically significant. Statistical analyses were performed using SPSS (SPSS 22.0, IBM Inc., Somers, NY, USA).

Post-hoc power analysis was performed to evaluate the power of this study. With a total sample size of 14 we achieved a statistical power for maximum load to failure of 90.9% (G*Power version 3.0.10; Franz Faul, Universität Kiel, Germany).

Results

The median failure load with monocortical fixation was 221 N, interquartile range (IQR) 211–297 N. The median failure load in the bicortical group was 423 N, IQR 273–497 N. The difference in the median failure load of both groups (202 N) was statistically significant (p = 0.017, Fig. 3).

Fig. 3
figure 3

Median failure load with IQR (p = 0.017)

Full size image

The mode of failure with monocortical fixation was a complete pullout of the screws in four cases (57%), and fracture of the glenoid in three cases (43%). The mode of failure with bicortical fixation was a complete pullout of the screws in two cases (29%, Fig. 4), and fracture of the glenoid in five cases (71%, Fig. 5). There were no statistically significant differences between the modes of failure (n.s.). In no cases did the construction fail at the clamp–bone interface.

Fig. 4
figure 4

Screw pullout as mode of failure

Full size image
Fig. 5
figure 5

Glenoid fracture as mode of failure (left scapula)

Full size image

Discussion

The main finding of this study was that monocortical fixation of the coracoid graft in the Latarjet procedure is significantly weaker than bicortical fixation in a biomechanical sawbone model.

The overall rate of postoperative complication after open Bristow or Latarjet procedures has been reported to be between 15 and 30% [14, 25]. Reported complications include neurological injuries, postoperative infections, bony non-union or graft osteolysis, screw bending, breakage or migration, and recurrent instability [10, 14, 15, 18, 26, 28, 29, 32, 37, 38].

Neurological injury is one of the most common complications after a Latarjet procedure. These injuries include brachial plexus traction injuries and axillary-, musculocutaneous- and suprascapular nerve palsies [11, 14, 28].

Screw positioning and length is critical because inaccuracy is associated with a higher rate of complications [2]. Studies have reported that 4.2–6% of the screws were too long [6, 21]. Hardy et al. reported that 63% of their screw tips were further than 2 mm away from the cortex of the posterior rim of the glenoid after arthroscopic Latarjet procedure [16]. This finding is supported by a cadaveric study which demonstrated that the screws fixing the coracoid bone graft were too long in 42% of cases [13]. This is problematic because of the proximity of the suprascapular nerve to the exit sites of the screws [4, 22, 28, 31]. In an anatomic cadaveric study, the mean distance between the major branch of the suprascapular nerve and the exit site of the superior screw was reported to be only 4 mm, with obvious contact between the screw and the nerve in 20% of cases [22].

Subsequent cadaveric studies have focused on defining a safe zone at the posterior rim of the glenoid to facilitate safe drilling and screw positioning. As the suprascapular nerve approximates the posterior border of the glenoid rim at a distance of 2.1 cm in the coronal plane, these studies have identified a 2 cm wide zone medial to the posterior glenoid, which is safe for the drill and the exit sites of the anteroposterior screws [13, 17, 35]. Furthermore, the drill angle required to avoid damaging the suprascapular nerve was found to be less than 28° of medial tilt in the transverse plane with respect to the glenoid joint surface, and greater than 29° from cranial to caudal in the sagittal plane [4, 34]. Bigliani et al. also concluded that anteroposterior screws directed inferiorly were the least likely to injure the suprascapular nerve [4]. Longo et al. reported that the suprascapular nerve is furthest away from the glenoid rim with the shoulder in 90° external rotation, and therefore recommends placements of the screws in this position during glenoid bone block procedures [24].

Yet short screws or screws that are not perpendicular to the osteotomy surface may compromise the stability of the coracoid fixation, and could lead to non-union and migration of the bone graft [5, 6]. The rate of coracoid process non-union or fibrous-union has been reported to be between 6 and 17% [8, 14, 19]. As intraoperative measurement of the optimal screw length can be difficult to achieve, preoperative CT planning to predict optimal screw length can be used to avoid screw length inaccuracy [16]. Furthermore, alternative fixation techniques are currently being studied to avoid screw mediated complications, such as cortical buttons [12].

Our findings suggest that better stability was achieved with bicortical fixation compared to unicortical fixation. Shin et al. compared monocortical fixation of the coracoid process with two 4.0 mm partially threaded solid cancellous screws in a fresh-frozen cadaveric model with bicortical fixation with the same type of screws and three further different types of screws [33]. In contrast to our study, they could not detect any statistically significant difference in load to failure or work to failure. That may be due to the different experimental set-up (sawbones vs. fresh-frozen cadavers), furthermore, they used 4.0 mm cancellous screws whereas we used 3.5 mm cortical screws. It may be the use of two cancellous monocortical screws instead of two cortical monocortical screws, as utilized in this study, may enhance the biomechanical stability.

The present study demonstrates that monocortical screw fixation reduces the fixation strength of the coracoid process by 45% compared to bicortical fixation. Further studies are needed to investigate if the inferior failure load of the monocortical fixation is clinically significant. A more controlled conservative postoperative rehabilitation program of the shoulder may mitigate the reduced stability compared to the classic bicortical fixation technique. However, in the absence of this, we recommend focusing on strategies to prevent SSN injury as outlined in Table 1.

Table 1 Current strategy to prevent injury to the suprascapular nerve in our clinic
Full size table

Strengths of the study were the homogeneity of the 4th generation Composite Scapulae. The influence of different bone quality on the stability of the construct did not have to be considered, as occurs in a cadaver study. We used a standardized technique for the operation of the Sawbones® and the biomechanical testing.

This study has the following limitations. The stability of the coracoid fixation only relates to a force applied directly to the coracoid bone graft as no experimental construct with an attached conjoined tendon was established. The load was applied perpendicular to the glenoid rim, and therefore not in the anatomic direction, as it is applied through the conjoined tendon in vivo. Further studies may involve using larger or cancellous screws.

According to this biomechanical study, clinicians have to consider that monocortical fixation of the coracoid minimises the risk of suprascapular nerve injury and soft tissue discomfort, but is less stable than bicortical fixation. Further research is necessary to evaluate whether this is clinically relevant and how much strength and stability is needed to achieve bony healing.

Conclusion

Bicortical screw fixation of the coracoid process graft in a Sawbone® model provides significantly stronger fixation than monocortical fixation. However, screw prominence may jeopardize the suprascapular nerve and it is not tested yet, how much stability is necessary to achieve bony healing. Anatomic knowledge of the safe zone at the posterior rim of the glenoid is crucial.