Use of cross-link in spinal fixation: a multi-body perspective to evaluate the loads interchanged in the implant

The spine is one of the most frequent site of cancer metastasis, impacting over 30% of cancer patients, mainly in the thoracic and lumbar regions. One of the most severe complications of spinal metastases is spinal cord compression, which may require surgical intervention to relieve neurological deficits. Decompressive laminectomy is a common solution, in which the vertebral laminae are removed through a posterolateral approach. This procedure causes strong spinal instability, requiring subsequent fixation of the affected vertebra, traditionally achieved through posterior multilevel fixation above and below the affected vertebra using pedicle screws and rods. However, this long-segment fixation is highly invasive and clinical reports are recording a growing incidence of mechanical failure. Short-segment fixations, involving only the levels adjacent to the affected vertebra, have been proposed for mild and moderate instabilities; notwithstanding this strategy reduces operative exposure and preserves a larger number of intervertebral joints, but there are substantial concerns regarding the reduced mechanical rigidity it may provide to the spine, especially in case of failure, because of the limited number of pedicle screws that could compensate for the resulting gap. A possible improvement of this less-invasive solution is the integration of a cross-link, now possible in oncologic surgery thanks to the recent introduction of radiotransparent carbon fiber reinforced PEEK material. Hence, this study aims to quantitively assess the impact of cross-link augmentation on short-segment fixation by evaluating the forces and the stresses inside the implant. In this framework, different configurations were compared, by varying the position of the cross-link and the shape of the short rods (straight and curved). The work recurs to multi-body dynamic simulations with flexible bodies based on modal theory. A pre-existing non-linear multi-body model of the T12-S1 spinal segment including all the passive elements (i.e., intervertebral discs, ligaments and facet joints) was used. Another version of the model representing a decompressed unstable condition was already available. From this, different postoperative outcomes were modeled, including long or short rods (160mm and 80mm length, respectively), straight and curved ( ρ: 440mm) rods, and above all, with the transverse cross-link connecting the two longitudinal rods in two different positions, cranial and caudal with respect to the decompressed site. Subsequently, the designed spine-implant constructs, were simulated in flexion, extension, lateral bending, and axial rotation, applying various loading conditions: pure moment, pure moment with 400N follower load, simulating in a simplified way the compressive load acting on the column by muscular activity, and finally pure moment combined with the distribution of the weight force acting in detail on the spine. The results were evaluated in terms of vertebral ROM, forces transmitted to the anchorages between rods and vertebrae, and distribution of Von Mises stresses on the rods. The numerical results indicated that long fixations are hyperstatic, while short ones allow for a more physiological range of motion. Furthermore, the use of the cross-link improved the rigidity of the short fixation, particularly in lateral bending and axial rotation, and limited mobilization in the event of pedicle screw loosening, especially when the cross-link is in the most cranial position.