Effects on hip stress following sacroiliac joint fixation: A finite element study.

Abstract For those patients who suffer from low back pain generated by the sacroiliac (SI) joint, fusion of the SI joint has proven to be an effective means of stabilizing it and reducing pain. Though it has shown promise, SI joint fusion raises clinical questions regarding its effect on neighboring joints such as the hip. As such, the purpose of this study was to determine the effects of SI joint fixation on the hip. A finite element spine‐sacroiliac‐hip (SSIH) model was developed and its functionality was verified against SI joint range of motion (ROM) and hip contact stress, respectively. The intact model was fixed in double leg stance at the distal femora, and loading was applied at the lumbar spine to simulate stance, flexion, extension, right and left lateral bending, and right and left axial rotation. Functionality was confirmed by measuring and comparing SI joint ROM and contact stress and area at the hip with data from the literature. Following verification of the intact SSIH model, both unilateral and bilateral SI joint fixation were modeled; hip contact stress and area were compared to the intact state. Average hip contact stress was ~2 MPa, with most motions resulting in changes less than 5% relative to intact; contact area changed less than 10% for any motion. Clinical significance: these results demonstrated that SI joint fixation with triangular titanium implants imparted little change in stress at the hip, which suggests that the risk of developing adjacent segment disease is likely low. Future clinical studies may be executed to confirm the results of this computational study.

ilia (often denoted as nutation and counter-nutation, respectively 3 ), followed by axial rotation and lateral bending. 4 The joint, along with its surrounding tissues, may be disrupted by factors such as pregnancy or trauma, or it may suffer degeneration due to reasons of spine surgery, leg length discrepancies, or abnormal biomechanics. 2,5 To stabilize the joint and reduce pain, the SI joint has been fixed via both open and minimally invasive procedures, 6 with solid triangular titanium implants being one such example of the latter. 4,7,8 Clinical studies have proven SI joint stabilization with solid triangular implants to be an effective means of reducing pain, increasing quality of life, and reducing opioid use, 9-12 while computational and experimental biomechanics studies have demonstrated a reduction in range of motion (ROM) following placement of these triangular implants. 13,14 While fusion of the SI joint has shown promise, the use of triangular titanium implants to stabilize and fuse the SI joint raises clinical questions about how such fusion affects nearby joints. The phenomenon of adjacent segment disease (ASD) is well known in the spine; fusion of spinal motion segments has been correlated with additional stress on adjacent levels 14,15 including the SI joint. 16 Using finite element analysis (FEA), Lindsey et al determined that stabilization of the SI joint via laterally-placed triangular implants resulted in at least a 50% reduction in ROM at the SI joint but less than a 5% change in ROM at the lumbar spine. Though long term effects were yet unknown, the investigation showed little change in motion at the spine due to SI fusion, thereby indicating a lower potential to develop ASD. 14 While this study investigated effects upon joints superior to the SI joint, research into the effects of SI joint fusion upon inferiorly located joints, specifically the hip, have yet to be explored.
The highly mobile hip joint is comprised of the acetabulum of the coxal bone and the femoral head; together their articulation, along with the surrounding soft tissues, facilitate both forward motion and simultaneous control of the body's center of gravity. 17 The load-bearing acetabulum and femoral head are mostly covered in hyaline cartilage of varying thicknesses and supported by the underlying subchondral and trabecular bony structures.
Thicker sections of cartilage occur along the superior aspects of the articulating surfaces since these areas are prone to higher pressures during weight-bearing. [17][18][19] Consequently, changes in the distribution of weight along the femoral head surface can lead to atypical stress concentration in the hip cartilage and subsequent degeneration. 18 The link between cartilage stress and degeneration and subsequent onset of disease like osteoarthritis 20,21 has prompted various studies to seek a better understanding of normal cartilage stress. [20][21][22] Specifically, Anderson et al tested a cadaveric hip by simulating the reaction forces attained during walking, ascending, and descending stairs (motions investigated in Bergmann et al's study 23 ) and measured the resulting contact pressures using pressure sensitive film. 22 These data were then used to validate a patient-specific, finite element model of contact pressure. 22 Similarly, Henak et al studied contact pressure in the hip in a series of patient-specific finite element models; each model was validated against a specimen, and additionally constitutive models and material definitions were evaluated. 21 In both studies, the authors highlighted the use of such models to further study hip mechanics and the impacts of abnormal stresses on the joint. 21,22 Because degeneration of the hip, specifically the cartilage, may lead to disease, it is important to understand possible causes that may be attributable to this degeneration. Therefore, the current research sought to characterize hip stresses before and after fixation of the SI joint and assess the changes in contact stress with relation to future degeneration at the hip joint. This characterization was accomplished via a multipart finite element (FE) investigation: (a) a previously-generated, intact lumbopelvic model was leveraged to validate a unilaterally-treated SI joint model, (b) a hip model was generated and its function was validated against contact stress and contact area data from the literature, and (c) the two were incorporated into a single spine-sacroiliac joint-hip (SSIH) model to evaluate contact stresses at the hip due to uni-and bilateral SI joint fixation following applied loading.

| METHODS
To characterize hip contact stresses following SI joint fusion, individual segment models (ie, lumbopelvic and hip) were first validated, and then these segments were combined into a single, intact SSIH model that was verified. Intact conditions were simulated prior to treated conditions, which incorporated uni-or bilateral placement of three triangular titanium implants. In the following sections, steps taken to create the finite element segment models are described, followed by those taken to develop the combined model. Metrics for each part of the study are discussed.

| FE segment model-Intact, treated SI joint, creation, and validation
With regard to the SI joint model, the current work leveraged a previously-validated lumbopelvic model. 24 This model is described in detail elsewhere, 24 but briefly, it consisted of an intact, female lumbar spine (L1-L5) with pelvis and fixed proximal femora whose functionality was validated against experimental SI joint range of motion (ROM) data.
To represent the treated SI joint model, the previously validated, intact pelvis model 24 was modified to incorporate a unilateral SI joint treatment. Specifically, unilateral fixation was accomplished by laterally placing three triangular, titanium implants (SI-BONE, Santa Clara, California) of lengths 60, 55, and 55 mm across a single SI joint. The implant's material property was that of Ti6Al4V (E = 115 GPa), a titanium alloy. The press-fit of the triangular implants was defined as an interference fit of 0.2 mm, and a nonlinear surface-surface contact with a frictional coefficient of 0.2 was assigned between the bone and implant.
To validate functionality of the treated SI joint, experimental conditions described by Lindsey et al 13  rotation, and then aligning these defined planes with the anatomical planes of the pelvis. 25 Further, a scaling step was executed to appropriately position the femora relative to the acetabulae. As the femora originated from a second cadaver's scan, they required scaling such that (a) femoral head and acetabulum overlap would be avoided and (b) joint space between the two bony features would fall within the range of reported data (female right and left hip joint spaces of 3.43 mm ± 0.40 mm and 3.48 mm ± 0.68 mm, respectively). 26 A sphere fit was performed on each femoral head and acetabulum, and the radii of the two fits were subtracted from one another to determine the joint space. Per the two scaling criteria mentioned earlier, this process was executed for scaling factors of 95% to 100% in 1% increments; 95% was determined to be the appropriate scaling factor (no overlap observed with right and left joint spaces of 3.46 and 3.36 mm, respectively).
Hypermesh 14 (Altair Engineering Inc., Troy, Michigan) was used to mesh the hip model; tetrahedral elements comprised the cancellous and cortical bone of the femora. A mesh convergence study was employed for the hip joint model in which seed sizes of 2, 3, 4, and 5 mm were applied to the femora, and functionality was assessed by monitoring the contact pressure during the simulated positions corresponding to walking, ascending stairs, and descending stairs.
To simulate these positions, the femur was oriented relative to the pelvis as per the angles defined in Bergmann et al's data. 23 Once the difference in contact pressure for any two consecutive seed sizes was within 2%, the coarser of the two meshes, in this case 3 mm, was selected and applied to the final femoral models.
Once the hip model was generated, finite element analysis of the hip was run in ABAQUS 6.14 (Dassault Systems, Providence, Rhode Island). Bony material properties were assigned per previous studies (Table 1). At the hip joint, a surface-surface soft contact was defined between the femoral head and acetabulum to represent the cartilage.
The soft contact was then assigned an exponential pressureoverclosure behavior with contact pressure and overclosure defined as 50 MPa and 0.5 mm, respectively, in which stress at the hip joint would increase as the distance between the two contact surfaces (ie, femoral head and acetabulum) decreased.
Hip joint functionality was validated against research conducted by Anderson et al 22 and Harris et al. 20 The experimental study simulated motions, specifically walking, ascending stairs, and descending stairs, 22 (Table 1). 31 The SSIH model was assembled in ABAQUS 6.14, and the final model contained a total of 767 694 elements.
To reiterate, the focus of this portion of the study was to confirm the combined model's functionality given new boundary conditions.

| Treated SSIH FE model-Creation and functional comparisons against intact
Following all validation and verification activities, the SSIH model was treated to assess the effect of SIJ stabilization on the hip joint relative to the intact condition, specifically the stresses at the hip joint before and after stabilization. Here, a treated model refers to the combined model described above with one or more stabilized SI joints as previously described (Figure 3). Loading and boundary conditions, simulated motions, and observed metrics for the treated models were the same as for the intact SSIH model. Regarding the unilateral treatments, SI joint range of motion as well as hip stress and area were evaluated at both the ipsilateral and contralateral joints.

| FE segment model-Treated SI joint
The ROMs of the left and right SI joints ranged from approximately 1 to 1.5 for all motions (ie, flexion/extension, lateral bending, axial rotation), which compared favorably with published variability for the unilaterally treated SI joint model 13 (Figure 4).

| FE segment model-Hip validation
Contact stress was found to be similar in comparison to the data presented by Anderson et al ( Table 2). Peak stresses were higher than those found in Anderson et al (

| SSIH FE model-Intact
SI joint range of motion for the current, double leg stance model ( Figure 5) was less than that reported for single leg stance. 24 Further, when compared to double leg stance data reported by Joukar et al, 24 ROM for the right and left SI joints of the current model were consistent. Average and peak contact stresses at the hip (Table 4, Figure 6) were found to be less than those in the walking, ascending, and descending stairs models, though they were similar in magnitude to average stresses reported in Harris et al for various activities. 20 Finally, contact areas were on the same order of magnitude as those found in the literature (Table 5).  peak contact stress for either the right or left hip, both increases and decreases were observed, though the increases in peak stress were all below 5% (Figure 9).

| SSIH FE model-Comparison, intact vs treated
Notably, contact area in stance was higher for the right hip joint in comparison to the left; this was true for the intact model as well as across all treatments (Table 8). Hip contact area for both left and right hip joints changed less than 10% when compared to the intact case for any motion with various motions resulting in no change in contact area after treatment (Table 9). Stress patterns were similar among different treatments with two regions of contact located generally anteriorly and posteriorly ( Figure 10).

| DISCUSSION
The current study sought to characterize stresses on the hip joint of a spine-sacroiliac-hip model before and after SI joint fixation to assess the potential development of ASD at the hip. To develop the SSIH, Similar to what previous studies have shown, 13  reported. 33,35 Following treatment, generally low changes (10% or less) in stress were observed at each hip joint ( Figure 9). Changes greater than 10% mostly accounted for decreases in stress, while a few instances accounted for increases; however, these increases equated to magnitudes totaling less than 2 MPa, appreciably lower than the previously reported peak stress magnitudes.
With regard to contact area, a slightly higher area was measured for the right hip than the left in stance and bilateral treatment. These differences suggest that the previously noted asymmetry was likely due to an anatomical difference between left and right sides of the pelvis. 36 Overall, changes in contact area relative to intact were less than 10% regardless of treatment; this was true for both right and left hip joints. Finally, contour plots depicting the contact area for a given motion before and after treatment were observed; stress patterns were considered reasonable when compared to the literature. 20,22 While definitive clinical commentary cannot be made regarding the development of hip degeneration following SI joint stabilization with triangular titanium implants, the above results suggest minimal changes in contact stress and area occur at the joint; and therefore, the potential for developing hip joint degeneration is likely low.
Some limitations are present in this study. First, the femora of the model originated from a different CT scan than the spine and pelvis.
The lumbopelvic model used in this study was sourced from an earlier study and the specimen did not include femora at the time of the scan; thus, femora were sourced from a second scan. However, to lessen the impact of the difference in specimens, the models were both generated from female donors. Further, to accommodate potential anatomical differences in fit between the acetabulae and femora of the two specimens, the femora were scaled and aligned relative to the pelvis per Wu et al. 25 Interference between the femoral head and acetabulum for each hip joint was assessed prior to finalizing the model such that the joint space was representative of data found in the literature. 26 A second limitation of the model was the representation of hip musculature as connector elements for the purpose of simplifying the overall model. Though the muscles were characterized by passive connector elements rather than active elements equating to muscle contraction, they sufficiently supported the combined model during the applied physiologic loading. Further, the connector elements were attached to the model at physiologic origin and insertion points, which ensured that they were acting at realistic locations on the model.
Despite the simplification, the model allowed hip motion and thus represented an improvement over other finite element models described in the literature, which included fixed acetabulae or immobile femora. 13,24,37 Future studies could expand upon the current model by incorporating active musculature as a physiologic constraint.
Finally, this study utilized double leg stance with a moment applied at the spine of a single model; other motions or loading configurations were not examined here. While "worst case" movements demonstrating impact such as running or jumping were not modeled, a stance model with applied moment was considered representative of day-to-day demands on the body. Further, while different test protocols exist, (ie, specifically load-control, displacement control, or a hybrid protocol), a load-control approach better demonstrates the post-operative motion of a patient in comparison to the displacement and hybrid control methods. 38 Additionally, in a previous study, little difference (6%) was demonstrated in the applied bending moment 14 when using the hybrid protocol to evaluate SI joint treatment via triangular titanium implants. As the difference in loading was minimal, a load control (moment with follower load) approach was adopted here.
Regardless of the test protocol, the current work presents a multisegment model with mobile hip joints that allows for the investigation and, moreover, quantification of hip contact stress and area, which is a new and useful contribution to the field. Future studies could expand upon this work by increasing the number of modeled specimens and/or modeling the hip joint in postures other than stance.
The current research demonstrated the effects of SI joint stabilization on the hip joint. Utilizing a previously developed model that combined the spine, pelvis, and femora, SI joint ranges of motion and stresses and contact area at the hip joint were evaluated following unilateral (left or right) and bilateral treatment of the SI joint with triangular titanium implants. Relative to the intact condition, SI joint ROMs were reasonable and changes in stress were mostly minimal; all stress magnitudes were within reported ranges for hip cartilage. Contact area also demonstrated small changes following SI joint fixation.
While computational studies alone cannot be used to make definitive clinical conclusions, the outcome of the current work suggests that the risk of developing adjacent segment disease at the hip joint following SI joint stabilization is likely low. Further studies may be conducted to assess patient outcomes and clinically confirm the results of this computational study.

ACKNOWLEDGMENT
Financial support for this research was provided by SI-BONE.