The first main finding of our study was that the mean coracoid bone graft length available for transfer (from the coracoid tip to where the inferior cortex curves inferiorly) in our young population was 23.9 ± 3 mm. Therefore, in 46% of the cases there was less than the ideal length of 25 mm, according to the original technique [11, 15]. Previous reports have shown divergent results regarding the maximal harvestable coracoid length. Dolan et al. and Shibata et al. reported in cadaver studies that the mean maximum length available for transfer was 28.5 mm [16] and 27 mm [17] respectively. Furthermore, Paladini et al. in computed tomography study and Young et al. in an intra-operative measurement study, found the available coracoid length to be 26.3 mm and 26.4 mm respectively [18]. However, Bhatia et al. and Lian et al. reported values of 19 mm and 24 mm respectively [19, 20]. From the aforementioned studies and the international literature, the coracoid length seems to be positively affected by male gender, age, overall height and Caucasian genetics [12, 13, 17, 20, 21]. In our study population, the height of the cadavers was not available, however we observed a positive correlation with the glenoid height.
The second main finding was that the harvestable coracoid length directly affects the position and possibly the biomechanics of the different implants used during fixation in the Latarjet procedure. Based on the AO principles we hypothesized that when the distance between the implant and the coracoid osteotomy is at least equal to the diameter of the implant, it would be “safe”, with low fracture risk. Therefore, complications such as acute coracoid fracture or later osteolysis could be avoided [10]. With the use of the originally proposed 4.5 mm malleolar screws this “safe distance” was ensured only in 56% of the cases. This percentage was significantly improved with smaller screws or the use of cortical button fixation. Despite this, no studies exist comparing the different types of fixation. This could explain the short-term differences in coracoid fracture and osteolysis when 4.5 mm screw fixation [22, 23] is compared to button fixation [24, 25]. However, when Boileau et al. perform their Latarjet technique, only 15 mm of the coracoid is harvested and fixation is with only one button [24]. In our study, for homogeneity reasons, we created a hypothetical model in which the complete coracoid is harvested and there are two points of fixation.
Furthermore, we found that the coracoid graft thickness is able to restore the glenoid anatomy in most of the cases when the classic Latarjet technique was performed. The mean “filling ratio” could be 3.3 to 1.7 in cases with smaller glenoid bone loss (10% to 20%) and 1.3 in most of the cases with defects of 25%. This extension of concavity of the glenoid articular arc may better manage the bipolar-“off-track” lesions [26] and explains the favorable clinical outcomes and the lower recurrence rates when the Latarjet procedure is performed in cases with even 13.5% glenoid bone loss [6, 27]. However, we found that in cases with 30% glenoid bone loss, the coracoid graft was not always enough to achieve a “filling ratio” of 1.0 (20% of the cases). Hantes et al. reported similar results in their cadaveric study. In intact glenoids with a mean area of 734 ± 89 mm2, they created a defect of 29%. After the reconstruction, the mean surface area of the glenoid was still smaller (708 ± 71 mm2), but this was not statistically significant [28]. Furthermore, the authors found that the coracoid thickness represents 27 ± 5% of the intact glenoid [28].
In a clinical study, Moon et al. operated on 44 patients with large glenoid defects of 25.3% ± 6% of the intact glenoid surface. Using 3-D CT-scans, they found 1.5 ± 2% recurrent bone defect, however, this did not affect the clinical outcome [29]. Also, Paladini et al., in 23 patients with glenoid defects greater than 20%, found that the coracoid filled the defect by 102% [18]. These small differences in filling the defect in published studies could be due to differences in age, race and patient height [13, 20]. However, when the glenoid bone loss is ≥30% the “congruent arc”- modified Latarjet [12] or the Eden-Hybinette procedure [7] could be preferred. Regarding the “congruent arc” technique, our results also showed that it may be indicated in greater defects, as the coracoid width was always greater that the coracoid thickness.
Young et al. performed the classic Latarjet technique with 4.5 mm malleolar screws, and during their intra-operative measurements, found that the distance from the edge of the inferior drill hole to the lateral margin of the graft was 5.7 ± 1.1 mm [30]. Sahu et al., also using 4.5 mm screws in their cadaveric study, found that this “lateral offset” of the coracoid graft was 5.5 ± 1 mm [31]. Our study results were similar, when using screw fixation. However, surgeons should be careful when the drill holes are placed in the middle of the coracoid and smaller implants are used. In these cases the “lateral offset” could be increased (by up to 9 mm).
Strengths and limitations
In this anthropometric study we tried to describe the possible implications between the coracoid process dimensions and the different surgical techniques. A strength of this study is the young male population (soldiers from the Second World War) that corresponds to the type of patients commonly operated on for shoulder instability. However, we cannot take into account the evolution of the Greek population over time, and nowadays larger glenoid and coracoid dimensions could be present within the population. Despite the measurements being reproducible (intra and inter-observer reliability), this was not a cadaveric study and the soft tissue insertions (pectoralis minor, coracoclavicular and acromioclavicular ligaments) were not present. Also, other anthropometric details of the specimens (like the height) were not available. The glenoid bone loss and the application of different materials were performed by using a “hypothetical” model and not in real practice. Finally, for homogeneity reasons all measurements were performed regarding the classic Latarjet technique and not the congruent arc or the Bristow technique.