Reproducibility of the anatomical model of TTE

These results allow us to answer our hypothesis. Indeed, among the 5 elbows of the two different protocols, ten TTE were created using the model: 9/10 (90%) with posterolateral and 1/10 (10%) with divergent anteroposterior dislocation.

The classic posterolateral dislocation in TTE has been reported previously [7, 8, 11]. Fitzpatrick & al. created 6 TTE with posterior dislocation under axial compression from 7 anatomical models positioned in pronation, but no randomization between models was performed [7]. Wake & al. created 11 TTE with posterior dislocation under axial compression from 15 elbows in full extension or 30° flexion; however, they did not detail pronation-supination and the models used formaldehyde-treated elbows, which would alter tissue properties and affect lesion onset [15].

Divergent dislocation was described radiographically by DeLee in 1981 [21], but is rarely seen in practice [22]. It includes 2 forms: anteroposterior divergence, which is the more common, and mediolateral divergence, less common [22,23,24].

In anteroposterior divergence, several studies have identified the mechanism as axial compression of the hand in extension, forearm in pronation and elbow in sub flexion, with a pivoting movement of the body relative to the hand. Thus, the lesion mechanism corresponded to the current test-bench position of the elbows, accounting for the creation of a single divergent anteroposterior TTE with probable error in fixation of the specimen on the test bench. Finally, the present study found 8 Regan-Morrey type-2 CP fractures and 2 type-1, in agreement with Doornberg & al., who reported that CP fracture in TTE occurred at approximatively 35% of CP height [25].

We reproduced TTE with the forearm in maximum pronation. This was chosen based on Fitzpatrick’s findings [7], with simple dislocation when the forearm was in maximal supination and associated RH fracture when in pronation. However, it is well described [10, 26] that the TTE lesion mechanism involves the elbow being in mild flexion and the forearm in supination. The present study casts doubt on this, as did Schreiber [17], whose video analysis found 70% of elbows in pronation at the time of trauma.

Loads to failure according to model

To study loads to failure, we did not position sensors on the structures involved by TTE, as this would have required at least slight opening of the joint capsule, which would undoubtedly alter the biomechanical properties of the model.

During axial compression of the whole elbow resulting in TTE with posterolateral dislocation, there was systematic valgus displacement and pathological forced external rotation (PFER) of the forearm. This corresponds exactly to the description by O’Driscoll & al. [8, 10, 18]. Also, the lesion chronology in TTE with posterolateral dislocation has also been described in several studies and although other theories have been proposed [1, 17, 27,28,29,30], the main one is the Horii circle described by O’Driscoll & al. and adopted by many authors. This begins with capsule-ligament rupture, first in the LCL and progressively extending laterally and medially, with possible injury to the posterior and transverse MCL bundles and, less commonly, the anterior bundle. This is followed by RH and CP passage fractures during the posterior dislocation [5,6,7,8,9,10]. The present elbow displacement patterns corresponded exactly to this pathophysiology as described by O’Driscoll, with 3 loads to failure seen on the load-displacement curves and lesion assessment systematically finding LCL tear associated with RH and CP fractures, without visible tear in the MCL bundles. Capsule rupture is progressive under these circumstances [5,6,7,8,9,10] and was thus not seen as an inflexion in the load-displacement curve.

This precise chronology thus sequences failure loading in each structure of the triad: first, LCL tear, then RH fracture, and finally CP fracture.

Mean loads to failure, for all elbows of the 2 protocols combined, in the LCL were 3,126 ± 1,066 N, in the RH 3,026 ± 1,308 N and in the CP 2,613 ± 1,120 N.

To the best of our knowledge, Fitzpatrick & al.’s study was the only one to report a mean load to failure in TTE, at 2,355.4 ± 339.8 N, without, however, detailing the number of loads to failure observed or their association with damaged structures [7].

Amis & al. reported mean loads to failure of 2,900 N (range, 300-6,100) in the RH and 4,300 N (range, 1,600-6,000) in the CP; however, their study concerned fracture outside of TTE contexts, with a direct or indirect trauma protocol very different from that used in the present study [12].

Since dislocation is a sudden event [20], the rapid protocol more closely approximated in-vivo trauma.

In rapid protocols in cadavers 1 to 4, comparison of loads to failure per structure (LCL, RH, CP) found no significant differences (p = 0.39), with equal p-values for each structure, likely due to small numbers, whereas comparison of mean loads to failure in all structures in a given elbow found a significant difference (p = 0.03). This may be explained by differences in age (78–98 years) and gender between cadavers.

In the light of these differences between cadavers, we compared differences in loads to failure per structure (LCL, RH and CP) between the slow and rapid protocols in a given cadaver (1 to 4). This found a low mean value of -4%, with random homogeneous distribution around zero. In this small sample of 4 cadavers, this suggested that there was no systematic over- or under-estimation of loads to failure between the 2 protocols for a given cadaver.

In the present series there were no lesions in the anterior, transverse or posterior MCL bundles in TTE with posterolateral dislocation. Although several studies did report MCL lesions in simple or complex elbow dislocation [28, 30, 31], O’Driscoll did not and explained this classic absence of anterior MCL bundle lesion in TTE [32, 33] by the RH and CP fractures, which allowed much of the dislocation energy to dissipate [10].

Relative forearm rotation with respect to the humerus prior to dislocation between models

Among the 5 elbows of the two different protocols, we found systematic external rotation of the forearm with respect to the humerus in all models under axial compression, with a mean 1.6 ± 1.2° prior to dislocation.

As described above, comparison of differential loads to failure per structure (LCL, RH, CP) between slow and rapid protocols in a given cadaver showed no systematic over- or under-estimation between the two. Thus, relative rotation values obtained with the slow protocol should be applicable to the rapid protocol.

Many authors have reported forearm PFER as the initial stage in posterior dislocation of the elbow [6, 11, 26]. It is defined as the combined relative rotation of the ulna with respect to the humeral trochlea and of the radius with respect to the capitulum, as distinct from pronation-supination defined as the relative rotation of the radius with respect to the ulna [7]. It is explained by the inclined surface of the lateral part of the medial two-thirds of the humeral trochlea, converting forearm axial compression force into lateral rotation [27].

However, PFER, does not appear to be systematic: Fitzpatrick & al. reported that 4 out of the 6 TTE they created started with PFER with initial LCL tear, and 2 with forced forearm internal rotation with initial MCL tear, and concluded that whether forearm rotation was lateral or medial determined whether the LCL or the MCL was torn first [7].

Study limitations and strengths

To our knowledge, this was the first study to create a reproducible anatomical model of TTE, and measure relative forearm rotation with respect to the humerus prior to dislocation and loads to failure in each structure involved by TTE. Previous studies focused on the theoretical, pathophysiological or therapeutic principles of TTE [8, 10, 16, 17, 28, 29], or on the biomechanical consequences of each individual lesion [5, 6, 13, 14], sometimes including anatomic models of elbow dislocation, but none have addressed the specific issues of TTE, and moreover all had serious limitations [7, 12, 15]. Another strength was the use of fresh specimens, without the formaldehyde used in other studies [15] which is known to alter the biomechanics of tissue by solidifying and increasing stiffness [34, 35]. Other strengths were that randomized matching between the slow and rapid protocols avoided numerous confounding factors, and the study was conducted in a laboratory certified and specializing in biomechanics. Finally, reproducibility between models was ensured by a precise dissection technique performed by a single surgeon, with standardized positioning on the test bench in constant 15° flexion and maximal pronation.

Nevertheless, 1 TTE showed divergent dislocation while the contralateral elbow from the same cadaver and the other 8 models showed posterolateral dislocation, suggesting that reproducibility could be further improved, as a lack of reproducibility in flexion may affect the lesion that is created. Wake & al. showed that, when elbow flexion increased under axial compression, the compression forces shifted from the CP toward the olecranon [15].

The other study limitations were the lack of bone density quantification by imaging which is inherent to any in-vitro study, making extrapolation to clinical settings.