The main result in the present study is that during flexion–extension movement with a load of 2 kg applied distally to the elbow, as occurs during minimal daily activity, a high stress on the bone and cement is generated. The greatest concentration of these stresses in the von Mises equivalent is localized in the distal part of the humeral component and in the most proximal part of the ulnar component. These are the areas where a high concentration of stress could cause implant failure if the applied force is prolonged [18]. Further, the displacement of the axis of the ulnar stem by −3° in the sagittal plane results in less stress, proving that variations in the positioning of the components with respect to the anatomical axis can lead to a potential improvement in biomechanics and therefore a longer predicted survival of the prosthesis.

There are various types of TEA, which, based on the connection between the humeral and ulnar components, can be categorized into constrained, semi-constrained, and unconstrained. Currently, most systems are semi-constrained. Therefore, they present a certain degree of laxity at the junction of the components. In the present study, the authors decided to use a constrained system in order to reduce the computational load of the FEM analysis and to evaluate the results as a whole rather than to consider the two semi-prostheses as separate entities. A choice was also made to cement the components, since, as stated by Fevang et al. [9], cemented prostheses have a higher survival rate than non-cemented prostheses.

The results of TEA have improved over the last few decades, and this can be attributed primarily to better biomechanical knowledge of joints and implants. TEA, moreover, is a technically demanding orthopedic procedure in which the precision of the restoration of the center of rotation of the implant as well as the correct positioning of the components are associated with better functional results, fewer complications, and therefore longer survival. However, high revision rates are reported, probably due to aseptic loosening from incorrect alignment between the axis of the constraint and the anatomical axis of rotation. In fact, due to aseptic loosening, rates of implants are reported to be between 47 and 77% with various types of prostheses [7]. Brownhill and Shuind have in fact shown in vitro that bad positioning of the ulnar or humeral component modifies the kinematics of the artificial joint and can determine its “loosening” [10, 11]. Few articles describe the effects of prosthetic implant placement on the stresses on the cement and bone for both the ulna and the humerus. Even less is known about the distribution of stress in this joint. Ericson et al. observed that the movement model of a TEA is, however, much less constrained than a normal elbow and this is more evident in unconstrained implants, with regard to the integrity of muscles and ligaments [12]. The humeral stem is typically valgus to fit the medullary canal; however, in some models, it is perpendicular to the hinge. Therefore, in terms of size and morphology, it must be well cemented, since the humeral canal tends to widen distally and has a thin cortex. On the other hand, the ulnar stem adapts well to the medullary canal of the ulna, which also has a thicker cortex. In 1986, Harry E. Figgie [13] described how small changes in the alignment of the implant with respect to the anatomical structure of the elbow have great effects on the functional results due to the high forces that develop in the joint during flexion and extension.

In a retrospective study on 25 patients operated on for TEA for rheumatoid arthritis or elbow fracture, Lenoir [14] evaluated the clinical outcomes, pain, and functionality of the prosthetic elbow in correlation with the correct positioning of the prosthetic components. Using computed tomography, the anterior offset, lateral offset, valgus, height, and rotation for the ulnar and humeral parts were examined. These indices provided a quantitative assessment of how position errors for the two components had additive or, conversely, counterbalanced effects on each other. The discrepancy between the humeral and ulnar lateral offsets was significantly associated with pain intensity and the Mayo Elbow Performance Score (MEPS); an anterior position of the ulna relative to the humerus was associated with reduced extension force and poorer outcomes for all functional parameters.

Even in the absence of implant loosening, positioning errors still seem to negatively affect the functional results, probably exerting inappropriate stress on the soft tissues. In Lenoir's study, it was also argued that neither the valgus index nor the rotation index is associated with clinical outcomes [14]. Ultimately, a slight posterior offset of the humeral component and a slight anterior offset of the ulnar component is recommended. In the frontal plane, the implants should be aligned with the native anatomical axes, as shown in Fig. 17.

Incorrect positioning (A) vs. correct positioning (B) of the prosthesis/bone axis

The main limitation of the present study is represented by the use of simplified mathematical models that may not be fully comparable to complex anatomical situations in vivo; also, it is not possible to relate the study to long-term implant survival. Further, our evaluation takes into account the most common stresses in flexion–extension and not in pronation–supination movements, so it is not possible to evaluate torsional stresses. However, it can be considered that recommending modest activity in daily life may reduce the risk of possible mobilization of the prosthetic components [15,16,17,18].