Model-based roentgen stereophotogrammetric analysis using elementary geometrical shape models: 10 years results of an uncemented acetabular cup component

The medical intervention for hip osteoarthritis finally ends with a surgical treatment of the affected joint by total hip arthroplasty (THA) [1]. Forecasts for primary and revision THA procedures predict an increasing case number [2]. However, THA presents an established orthopaedic intervention with a reported survivorship greater than 95% at 10-year and 80% at 25-year follow-up [3]. National registries [4] and clinical studies [5, 6] indicate that THA survivorship is affected by several factors like total joint arthroplasty (TJA) design, bearing couple or surface texture of non-cemented designs. In summary, THA failure most often is caused by aseptic loosening [3, 4, 7] and periprosthetic joint infection [8, 9], both associated with an important economic impact [10].

According THA, the interpretation of survivorship rates must be performed separately for femoral stem and acetabular cup components. For instance Hallan et al. [11] reported a good survivorship at 15-years follow-up for non-cemented femoral stems, whereas non-cemented metal-backed acetabular cup components revealed poor clinical outcome in comparison to cemented all-polyethylene components. Non-cemented metal backed cups showed different clinical outcomes, i.e. grit-blasted and hydroxyapatite-coated acetabular cups indicate a three times higher risk of revision than hydroxyapatite-coated, threaded acetabular cups in cause of aseptic loosening [11]. Palomäki et al. [12] reported data for uncemented cups of the second generation, collected within the Finnish Arthroplasty Register, showing survival rates similar to cemented treatment. However, these results based on mid-term data, longtime data for these cups were still necessary.

The clinical performance of non-cemented coated TJA components in general depends on several factors predicting the bony osseointegration: (i) properties of the implant surface likewise surface texture, coating layer thickness and applied pore size, (ii) host bone quality, (iii) surgical site preparation, (iv) preoperative loading conditions and (v) preventing initial and chronic infections [13, 14].

If the bony osseointegration of a TJA design fails, loosening of the implant will occur. Gold standard method to assess the in vivo fixation of TJA components presents Roentgen Stereophotogrammetric Analysis (RSA) [15, 16]. It is important to monitor TJA fixation within preclinical trials using gold standard methods, since worst case scenarios of clinical TJA failure – likewise the Boneloc cement disaster within the early 1990s – could be avoided [17‐20] and the migration pattern of TJA is affected by the implant design and its fixation philosophy [21, 22]. Using RSA, early implant migration within the first two postoperative years could be detected, which has been shown to correlate well with later aseptic loosening of femoral stems [23] and acetabular cups [24, 25].

Several variations of RSA exist, i.e. using markers attached to the implant (marker-based RSA) or a surface model of the investigated TJA components (model-based RSA) to calculate early implant migration [20, 26]. The interchangeable applicability of the both methods has been demonstrated [27‐29]. Model-based RSA using elementary geometrical shape (EGS-RSA) is an approach which uses geometrical bodies likewise spheres or cylinders to calculate migration by rigid body kinematics instead of implant specific surface models [30‐33].

Nieuwenhuijse et al. [24] identified thresholds for micromotion of a cemented acetabular cup to distinguish between safe or at risk implants. This research group found cranial migration > 1.76 mm or sagittal rotational migration > 2.53 deg. within the first two postoperative years to be risk factors for subsequent aseptic component failure of acetabular cups. Although a cemented cup was investigated, the authors state that the results of their study should be independent of the fixation method [24]. In accordance with this statement, Pijls et al. [25] reported in a systematic review a similar threshold for cemented as well as uncemented cups. A mean cranial migration of 0.2 mm or less was classified as acceptable, with revision rates of < 5% after 10 years.

Retrospectively, out of a former classical marker-based RSA study of 50 patients (Fig. 1a), RSA radiographs were analysed using model-based RSA approach with a hemisphere elementary geometrical shape surface model (Fig. 1b). Marker-based migration calculation was not possible due to implant-marker occlusion problems (in 43 of 50 cases) within the RSA radiographs (Fig. 1c, d). But using a model-based RSA approach with EGS models, migration calculation was possible in 39 of 50 cases.

Accuracy determined by using a phantom investigation showed better values for the marker-based RSA than for the model-based EGS-RSA method. However, both methods provided results within the range of RSA accuracy reported by Kärrholm [37], which is between 0.05 and 0.5 mm for translational migration. Thus, we consider that the loss of accuracy from using model-based EGS-RSA is acceptable for a clinical application of the methodology.

The clinical outcome of the investigated cup component indicates statistically significant migration according to the reference examination along the cranio-caudal axis at the 6 months (0.16 mm; p = 0.029), 12 months (0.16 mm; p = 0.029) and 24 months follow-up (0.17 mm; p = 0.017). At 120 months follow up, migration of the cup component increased significantly up to 0.36 (±0.27) mm (p < 0.001). It is important to note that in the first 24 months the observable maximum mean migration in cranial (0.11 mm) direction occurred within the first 6 weeks after surgery (Table 2). Afterwards a steady state mode of the detected migration was observable for the upcoming follow-up intervals within the first 5 years. Between five and 10 years the cup continued to migrate in cranial direction again by another 0.17 mm.

Two patients showed especially noticeable migration greater than 1.0 mm in the medial direction (Pat 1; 1.7 mm) and of approximately 1.0 mm in lateral direction (Pat 2) within the first 6 weeks, after which the cup appeared to stabilise up to the 24 months follow up (Fig. 4a). However, Pat 2 cup starts to migrate again in lateral direction after 24 months follow up. Conventional anterior-posterior radiographs indicate no clinical signs of loosening (e.g. radiolucent lines).

Some patients (n = 13 at 24 months; n = 11 at 120 months, including Pat 1) showed migration between 0.2 mm and 1.0 mm in cranial direction (Fig. 4b), however no cups were identified to migrate more than 1.0 mm. The maximum cranial migration value was observable after 6 weeks with 0.76 mm (Pat 1), which remained unchanged for this patient during the next follow-up investigations. Thus, this cup seems to have stabilised with no further migration in this direction.

In anterior-posterior direction one patient (Pat 3) had a migration value of − 1.28 mm at 6 weeks follow-up, which increased to > − 2.0 mm at the 24 months follow up (Fig. 4c).

Cranial migration is seen as a relevant predictive value for late aseptic loosening of acetabular cup components [24, 25, 38, 39]. In a systematic review, Pijls et al. [25] reported an average increase of 10% in the 10-year revision rate for every millimetre of cranial migration of an acetabular cup component. The authors suggested that inferior components could be identified by RSA within the first two postoperative years, thereby avoiding their widespread implementation. Furthermore, they provided proximal migration thresholds for acetabular cups, whereas cups with a mean cranial migration of 0.2 mm or less were classified as acceptable, with revision rates of < 5% after 10 years. In contrast, cups with a mean cranial migration of 1.0 mm or more were classified as unacceptable, with a > 5% risk for 10-year revision.

For the investigated cup component, a mean proximal migration of 0.17 mm occurred after two postoperative years (n = 36 patients at 24 months), which is less than the proposed threshold of 0.2 mm suggested by Pijls et al. [25]. In summary, 13 patients at 24 months were classified at risk (migration between 0.2 mm and 1.0 mm), no patient could be classified as unacceptable (migration > 1.0 mm). These results indicate good performance for the investigated cup component, according to suggestion of Pijls et al. [25]. As none of the followed implants had to be revised after 10 years, this study verifies the suggestions of Pijls et al. [25].

One limitation of this study was the absence of rotational measurements about the cranio-caudal axis. Due to technical restrictions of EGS-RSA, analysis of this rotational migration could not be conducted as the cup has an axis-symmetrical design. This is a well-known problem in model-based RSA [40]. Nonetheless, migration in cranial direction and rotation about the anterior-posterior axis could be calculated, which are the essential movements to predict later cup loosening [24, 25]. Current studies evaluate computed tomography (CT) as an alternative measurement tool to RSA to analyze implant migration. They describe comparable accuracy and precision values for CT-based methods, especially in experimental settings [41‐44]. Moreover, CT-based migration measurement could handle some disadvantages of RSA: CT is available in almost every hospital, picture acquisition does not require a special setup or briefed personnel, migration in all directions can be calculated, and there is no problem of marker occlusion by the implant [45]. On the other hand, conventional CT examinations expose patients to higher radiation doses than stereoradiographs, but there is some evidence that a reduction of CT radiation dose does not impair the accuracy of migration analysis [42, 43]. However, there is still a lack of clinical trials and long-term examinations evaluating CT-based migration measurement. Besides, there are still some problems to deal with, as for example artifacts of metallic implants, the influence of soft tissue or the immobilization of patients during CT scanning [44]. In summary, CT-based methods need further validation before they can be considered as an alternative to gold standard RSA.

Another limitation of this study is the number of drop-outs. However, the drop-out rate is comparable to those of other RSA studies [23, 24, 46‐48]. During the crucial period of the first 2 years, that are relevant for RSA-prognosis, eleven patients had to be excluded. The remaining number of 39 patients also complies with the RSA guidelines proposed by Valstar et al., who recommend a number of 15–25 patients for the special case of RSA studies [34].

Unfortunately, it is not possible to define prognostic migration thresholds out of the results, as there was no implant failure observable in 10 years of follow-up. However, the results confirm the above-mentioned thresholds of previous studies.

The migration pattern of the investigated acetabular cup component showed migration along all the three axes, particularly in medio-lateral and cranio-caudal direction. Migration along the latter mentioned direction is considered to be the most important prognostic factor for a later aseptic loosening but was still within acceptable limits according to Pijls et al. [25]. It would be interesting to follow the patients for a longer period in order to evaluate how the migration pattern of all axes will influence the long-term stability of the investigated non-cemented acetabular cup.

Model-based EGS-RSA enables an in vivo migration measurement of TJA components without the necessity of TJA specific surface models. The investigated acetabular cup component demonstrates significant differences for calculated migration values along the most predictable axis of acetabular cup loosening, migration along cranio-caudal axis with 0.17 (±0.24) mm at 24 months. However, the measured in vivo cup migration at 2 years of follow-up were within the acceptable thresholds reported in literature [24, 25]. For the cranio-caudal axis, migration continuously increased up to 0.36 (±0.27) mm at 120 months follow-up but is still within above mentioned thresholds. However, performed migration measurement by model-based EGS-RSA within this study highlights the importance and benefit of RSA for the evaluation of implants. There may be other possible applications in the future, such as the early detection of septic loosening of joint implants [49].