Treatment of Recurrent Dislocation after Total Hip Arthroplasty Using Advanced Imaging and Three-Dimensional Modeling Techniques: A Case Series

Dislocation is one of the most common complications seen after total hip arthroplasty (THA). It can occur during the early or late post-operative period [6]. Reported rates of dislocation range from 0.1 to 9% after primary THA and 5 to 30% after revision THA [1, 16, 17, 24, 37, 39]. Single episodes of instability may be successfully treated with reinforcement of “hip precautions” (exercises and activities to avoid) if the components are properly aligned and proper hip mechanics have been restored. However, in approximately a third of patients with dislocation, conservative treatments fail and surgery is required [14]. Recurrent dislocation after THA can be devastating and is the most common reason for revision THA in the USA, accounting for approximately 23% of revision THAs performed [9]. Unfortunately, surgical intervention does not always improve hip stability, and there remains a 21 to 30% risk of recurrent hip dislocation after revision THA [8, 12].

Surgical procedures commonly used to treat hip instability include increasing femoral head size; correcting malpositioned components; using an elevated liner, a dual-mobility construct, or a constrained liner; and repairing the soft tissues [28, 34]. The literature shows varying effectiveness of strategies for treating dislocation. A larger-diameter femoral head reduces the risk of dislocation caused by greater jumping distance and a greater range of motion before impingement [7, 27, 41]. But possible drawbacks of a larger head size in polyethylene liners might be greater wear of the liner and greater taper corrosion [15, 35]. Constrained or tripolar cup designs have reduced post-operative dislocation rates, but mechanical failure of the locking ring or dissociation of the cemented liner continues to be a problem [44]. The use of dual-mobility constructs has led to a clear improvement in terms of preventing dislocation, reducing the dislocation rate to 4% in revision THA after 6 months, but whether their ability to provide long-term stability or fixation longevity remains unknown [39].

Developing a plan for surgical treatment of dislocation may be difficult if the causes of instability are unclear. Dislocation after THA is thought to be related to impingement, a mechanical abutment between bone, implants, or soft tissues. Impingement is a dynamic process that is difficult to identify or characterize on the basis of clinical evaluation or plain radiographs [32] and may be driven by multiple factors, including hip offset, implant design, component position, and bony geometry. An understanding of the underlying dislocation mechanism is crucial to determining the appropriate surgical treatment for the instability [2, 13, 14, 23, 43].

An algorithmic approach to identifying types of impingement in THA may direct whether component revision or bony resection would be more effective in improving hip range of motion during revision surgery for instability. We used advanced imaging and three-dimensional modeling techniques to identify the type of impingement (bone on bone or implant on implant) occurring during simulated dislocation activities in patients undergoing revision surgery for instability to guide the development of a pre-operative plan to improve hip range of motion and treat instability. This series of patients was followed for 2 years after revision surgery to determine whether the treatment plan improved hip stability in the short term.

We obtained institutional review board approval to conduct a case series. We identified eight patients (two men, six women; mean age, 62 years) who experienced recurrent dislocation after THA from 2013 to 2015 (Table 1). The selection criteria were patients who required revision surgery for dislocation by two surgeons (D.J.M. or S.A.J.) during the 2-year period. All original THAs were performed using the posterior approach, and all hip dislocations occurred posteriorly as patients rose from a low chair, tied shoes, or bent over to reach objects. We had developed an algorithmic approach to pre-operatively identify the type of impingement occurring in each patient and to develop a treatment plan for each patient (Fig. 1). We built three-dimensional bone and implant models from advanced imaging and measurements of acetabular and femoral implant position (Fig. 2). Each patient underwent a computed tomographic (CT) scan, as well as standing biplanar frontal and lateral plane two-dimensional radiographs from the spine to the ankles using a low-dose radiation system (EOS Imaging System, EOS Imaging, SA., Paris, France). As the planning algorithm matured, we added sitting biplanar radiographs to account for pelvic alignment in different functional positions. CT scans were taken supine and included the pelvis from the anterior superior iliac spines to the proximal third of the femur, as well as the distal femur, in order to measure femoral torsion. The CT scans were segmented using MIMICs software (Materialise, Leuven, Belgium), and segmentation data from the pelvis, proximal femur, distal femur, femoral component, and acetabular component were exported as .stl files. We then aligned the models to functional imaging (standing, sitting, and supine radiographs) to simulate range of motion and measure component position in functional positions (Figs. 3 and 4).

We simulated the patient activities during which the hips dislocated to determine range of motion to impingement. To do so, the CT files were imported into a multibody dynamic modeling software (SimWise 4D, Design Simulation Technologies, Canton, MI, USA) for range of motion analysis (Fig. 5). Because the posterior dislocations occurred with flexion, adduction, and internal rotation, we calculated the range of motion to maximum flexion (with neutral abduction and neutral rotation) and maximum internal rotation at 90° of flexion (with neutral abduction). In the model, the pelvis was fixed in position, and the femur was rotated about the center of the femoral head in the motions described above. Contact conditions were established, so the analysis ended when contact was detected. To determine the threshold values for the acceptable or limited range of motion using our modeling methodology, we had previously measured maximum hip flexion, internal rotation at 90° of flexion, and external rotation at 20° of extension in seven fresh-frozen cadaveric pelvis-to-knee specimens from donors who had undergone THA. We found mean hip flexion to be 118° ± 6°, mean internal rotation to be 38° ± 3°, and mean external rotation to be 28° ± 5°. Two standard deviations below the means were 106° for flexion, 32° for internal rotation, and 20° for external rotation, which were the values we used to identify limited hip range of motion.

The location and type of impingement (bone on bone or implant on implant) were determined and used to develop a pre-operative plan for improving range of motion. Surgical options for improving hip range of motion included reorientation of the acetabular component, reorientation of the femoral component, revision of the femoral head to increase hip offset, and removal of impinging bone (Fig. 1).

The images shown in Figs. 4, 5, 6, 7, and 8 are representative of the pre-operative planning algorithm, although the patient, a 51-year-old woman undergoing revision surgery for hip dislocation, was not included in our study. Figure 5 shows how different surgical plans improve hip range of motion. Before treatment, the patient had limited range of motion in flexion and internal rotation as a result of bone-on-bone impingement between the anterior inferior iliac spine (AIIS) and the anterior aspect of the proximal femur; there was evidence of a prominent AIIS and a proximal femoral osteophyte, so the pre-operative plan was to remove bone at these locations (Figs. 6 and 7). The model showed that this plan would improve the patient’s range of motion from 105 to 113° of flexion and from 18 to 32° of internal rotation (Fig. 5). If the surgeon also elected to increase the neck length an additional 4 mm with the use of a high-offset head or lateralized liner, the hip range of motion would improve further, to 114° of flexion and 38° of internal rotation (Figs. 5 and 8).

In our study, revisions were performed through the posterior approach, and patients were followed for 2 years to determine outcomes.

The eight patients in this case series who experienced recurrent dislocations had variability in component orientation, hip range of motion, and hip impingement mechanisms (Table 2). Four out of the eight patients had acetabular components within the “Lewinnek safe zone” (40° ± 10° of inclination and 15° ± 10° of anteversion) when they dislocated, which indicates that this traditional safe zone does not provide a low risk of dislocation for every patient.

In four patients (patients 1, 4, 5, and 7), hip range of motion was limited in flexion (less than 106°) or internal rotation (less than 32°); in three of these patients (patients 1, 4, and 5), there was bone-on-bone impingement involving either the AIIS or an acetabular osteophyte, and in two patients (patients 4 and 5), the femoral component was in retroversion (Table 2). Patient 4 had a large AIIS that required resection of 1 cm of bone in order to prevent future anterior impingement on the femur. Patient 1 not only had a prominent AIIS but also had a proximal femoral osteophyte, which limited range of motion at the hip to only 83° of maximum flexion (Table 2). The osteophyte was removed intra-operatively. Figures 9 and 10 show a 69-year-old woman (patient 5) with a prominent AIIS and a cemented femoral component in retroversion. The AIIS and the position of the femoral component were not apparent on the conventional radiographs (Fig. 9). The surgical plan for this patient (Fig. 10) was to remove AIIS bone and to reorient the cemented femoral component with greater anteversion.

The other four patients in this study (patients 2, 3, 6, and 8) had acceptable ranges of motion (Table 2). This suggested possible soft-tissue causes of dislocation. In these patients, the surgical plan was to revise the acetabular component and implant either a dual-mobility bearing or an elevated liner (Table 3).

In summary, all but one patient underwent an acetabular component revision, and two patients underwent a femoral component revision for the treatment of recurrent dislocation in this study. No patients in this case series experienced a dislocation within 2 years after revision surgery.

Impingement involving bone, implant, or soft tissues is an important consideration in the surgical treatment of instability after THA [3, 4, 10, 33, 36]. Modeling hip kinematics may elucidate the underlying impingement mechanism responsible for hip dislocation. We considered implant position, hip range of motion, and location of impingement when developing a strategy for treating hip instability. In four patients, we found limited hip range of motion attributed to bone or implant impingement. However, in the other four patients, there was no evidence of limited hip range of motion when we considered bone and implants alone, suggesting soft-tissue causes of dislocation. To our knowledge, this is the first study to use a dynamic modeling tool for pre-operative planning of revision THA.

There are limitations to our study design. First, this study applied the treatment algorithm only to eight patients with hip instability; however, it is difficult to collect a large group of patients with hip dislocations because the dislocation rate after primary THA is low (1 to 3%). Second, although we used advanced imaging pre-operatively to develop treatment plans, we did not do so post-operatively to confirm surgical execution. We cannot be certain of how implants were reoriented or how much bone was removed during revision surgery. Third, we did not include pelvic tilt or leg length as a part of our algorithm because that would have necessitated a more complex and time-intensive analysis. Because these are important variables, we hope future models will automatically incorporate different pelvic tilt positions and leg lengths into the algorithm. Fourth, as a case series, our study may have involved selection bias; also, because it did not have a control group, the generalizability of our findings may be limited. Finally, we did not take soft tissues into account in our modeling efforts, but it is clear that soft-tissue repair is an important part of THA stability [38].

Bony impingement was common in our modeling analysis and may not be detected by THA surgeons because prominent bony features such as the AIIS cannot be seen on conventional anteroposterior radiographs. It is important to remember the principles regarding impingement in the native hip put forth by Ganz and colleagues [5, 22, 42]. An underlying bone-on-bone impingement in a patient’s native osteoarthritic hip may continue after THA [32]. AIIS deformity has been shown to be an extra-articular source of impingement in the native hip [25], and hypertrophy of the AIIS has been shown to limit hip range of motion [26]. Our models suggest mechanical abutment can occur between the AIIS and the femoral bone in patients with unstable THAs. In our study, bone-on-bone impingement occurred most commonly in hips with decreased anteversion of the femoral stem (less than 5°) and short neck length with the distal end of the AIIS and anterior superior aspect of the capsule often having to be excised.

Precision is important when placing THA implants because implant malalignment has been associated with instability, high wear, and poor hip range of motion [18, 32, 40]. We identified implant malalignment in hips with limited motion as a result of implant impingement. However, it was difficult to determine what the new alignment target needed to be for a malpositioned component, and the intra-operative execution of the pre-operative plan was not always easy. Component revision can be difficult, and it is not always in the best interest of the patient to remove a well-fixed implant [11]. For example, in patient 4, the femoral stem was found to be in 2° of retroversion, which was an indication for revising the femoral component (Table 2). However, the surgeon did not do so because he performed an intra-operative hip range of motion check and found no evidence of impingement after the cup was revised and the AIIS was resected. Rather than revising components, surgeons may elect to be conservative and increase neck length to improve hip range of motion. In all but one patient in this series (patient 7), a dual-mobility bearing or elevated liner was implanted for improved hip stability. Although these constructs may contribute to the favorable outcomes we report in this series, our models showed that merely exchanging a bearing for a dual-mobility construct may not solve a bone-on-bone impingement problem.

Accurate pre-operative assessment of implant position and impingement is dependent on the functional orientation of the pelvis and the femur during the performance of activities of daily living and the assumption of provocative positions that cause instability [30]. We used functional imaging in both the standing and sitting positions to align the pelvis and the femur in our models. In most patients, the pelvic tilt and femoral rotation in the CT scan differed depending on whether the patient was standing or seated (Fig. 3). Pelvic tilt will affect range of motion to impingement and is directed by the spine mechanics [20, 29]. Interestingly, all patients in this study had cervical, thoracic, or lumbar spine disease (or a combination of these). Two patients in our study had large thoracic scoliosis curves (Figs. 3 and 9). Spine disease can limit a patient’s ability to accommodate postural changes through the lumbar spine, which alters hip kinematics and increases the risk of hip dislocation [21].

The pre-operative planning performed in these cases required extensive communication between the modeling analyst and the orthopedic surgeon. Impingement modeling may show improved range of motion with component revision, but the orientation of the implants or the bony resections performed intra-operatively should still remain within an acceptable clinical range. Traditionally, the Lewinnek safe zone has been used as an indicator of a low risk of hip dislocation [31]. Interestingly, half of the patients in our study had acetabular components within the safe zone, and they still experienced hip dislocation. This supports evidence in more recent research showing that a truly safe zone for acetabular component position alone does not exist [19]. One incentive for surgeons to consider using this time-intensive pre-operative modeling algorithm is to avoid constrained liners, which may be beneficial in patients who have either instability of unclear etiology or cognitive problems but may not be ideal for high-demand (more active) patients requiring revision THA [44]. All patients in this study experienced posterior hip dislocations. However, by evaluating hip external rotation range of motion in extension, a similar algorithm could be used to determine treatment in patients with anterior hip dislocations. In future studies, this kinematic modeling platform can be used to plan optimal implant position or bony resections around native hips or THAs.