The Reconstruction Accuracy of the Hip Center of Rotation After Cementless Total Hip Arthroplasty for Failed Treatment of Acetabular Fractures. Does It Matter?

To assess the impaction of reconstruction accuracy of hip center of rotation (COR) on midterm clinical and radiographic results of cementless reconstruction of total hip arthroplasties (THAs) for patients after failed treatment of acetabular fractures. Methods One hundred and four patients (107 hips) who underwent THAs after failed treatment of acetabular fractures were retrospectively evaluated and cementless cups and stems were implanted in all hips. Clinical outcomes were assessed using the Harris hip score (HHS) and Western Ontario and McMaster Universities Arthritis Index (WOMAC) scoring system. Radiographic results were analyzed by serial perioperative x-rays.


Abstract Background
To assess the impaction of reconstruction accuracy of hip center of rotation (COR) on midterm clinical and radiographic results of cementless reconstruction of total hip arthroplasties (THAs) for patients after failed treatment of acetabular fractures.

Methods
One hundred and four patients (107 hips) who underwent THAs after failed treatment of acetabular fractures were retrospectively evaluated and cementless cups and stems were implanted in all hips. Clinical outcomes were assessed using the Harris hip score (HHS) and Western Ontario and McMaster Universities Arthritis Index (WOMAC) scoring system. Radiographic results were analyzed by serial perioperative x-rays.
Compared with normal contralateral hip, 79 cups migrated superiorly (0.2-33.6mm) and 22 cups migrated inferiorly (0. 2-16.1mm). The distance of superior migration of reconstructed COR was correlated with positive Trendelenburg sign at the latest follow-up examination (r=0.504; p < 0.001). The percentage of postoperative Trendelenburg sign was signi cantly higher in superior migration subgroup than that in subgroup with anatomical restoration of COR (P=0.015).

Conclusions
Cementless THAs in patients after failed treatment for acetabular fractures achieved predictable clinical and radiographic outcomes. A superiorly migrated hip COR appeared to exert a negative effect on abductor muscle function.

Background
Optimal restoration of hip center of rotation (COR) is of great importance for implant longevity in total hip arthroplasty (THA). Any errors in reconstruction of hip COR can inevitably affect implant load, stability, wear rates and can eventually result in poor hip function with costly complications, including pain, loosening and dislocation [1,2,3,4,5]. Acetabular fractures may lead to late complications, including posttraumatic arthritis and avascular necrosis of femoral head with or without initial open reduction and internal xation (ORIF) [6,7]. THA has been utilized to treat patients with disabling pain and functional limitations after failed treatment of acetabular fractures. Previous literature has demonstrated encouraging outcomes with regard to cementless acetabular components [8,9,10,11,12,13]. Among these, previous articles ever described the reconstructive accuracy of hip COR after THA, but failed to compare it with clinical results [8, 9,10].
The purpose of this study was to retrospectively review a single center's experiences in treating failed acetabular fractures with THA by cementless components. An additional purpose was to assess whether the reconstructed hip COR affects clinical results and complication rate.

Methods
Between 2001 and 2017, 124 consecutive patients were treated at our institute with primary THAs after failed treatment of acetabular fractures. The median duration of follow-up was 93.0 months (47-151 months) in 104 patients (107 hips) ( Table 1); 20 patients were lost to monitoring. Approval of the local institutional review board and informed consent from all participating patients were obtained. The initial fracture pattern was classi ed according to Judet classi cation system (Table 2). We exposed all 107 hips through a posterolateral approach. Hardware was partially removed in 21 hips, entirely in 26 hips, and left in situ in 20 hips. We implanted tantalum trabecular metal (TM) (Zimmer, Warsaw) cups in 38 hips and titanium hemispheric cups in the other 69 hips. In 69 cases, the cups were supplemented with 2 to 5 screws and the initial stability and orientation of the cup was con rmed with intraoperative radiographs. Cementless stems were implanted in all hips. We used a ceramic-on-ceramic bearing surface in 38 hips (35.5%), a ceramic-on-polyethylene bearing surface in 45 hips (42.1%), and a cobalt-chrome head on a polyethylene bearing surface in 24 hips (22.4%).  The initial fracture pattern was classi ed according to Judet classi cation system ( Table 2). We exposed all 107 hips through a posterolateral approach. Hardware was partially removed in 21 hips, entirely in 26 hips, and left in situ in 20 hips. We implanted tantalum trabecular metal (TM) (Zimmer, Warsaw) cups in 38 hips and titanium hemispheric cups in the other 69 hips. In 69 cases, the cups were supplemented with 2 to 5 screws and the initial stability and orientation of the cup was con rmed with intraoperative radiographs. Cementless stems were implanted in all hips. We used a ceramic-on-ceramic bearing surface in 38 hips (35.5%), a ceramicon-polyethylene bearing surface in 45 hips (42.1%), and a cobalt-chrome head on a polyethylene bearing surface in 24 hips (22.4%).

Clinical And Radiographic Evaluation
All clinical evaluations were conducted by independent observers (C.M.X., L.W.Y.) who had not participated in the THAs. All patients were routinely examined at three months, six months, one year, yearly until 5 years after surgery and then every 5 years thereafter. For patients who did not comply with routine follow-up schedules, data collection was accomplished via telephone, E-mail, or WeChat software. Clinical outcomes were evaluated using the Harris hip score (HHS) and Western Ontario and McMaster Universities (WOMAC)Arthritis Index scoring system. Patient satisfaction was also evaluated using a self-administered four-category scale (very satis ed, somewhat satis ed, somewhat dissatis ed, and very dissatis ed).
Serial radiographs included anteroposterior (AP), lateral, and two 45° oblique views of the involved hip before THA and at the follow-up visit. Computed tomography scans of the involved pelvis were routinely reviewed preoperatively. Radiographic assessments of acetabular components were accomplished in three zones, as devised by DeLee and Charnley [15]. Radiographic failure was determined by >3 mm of vertical or horizontal migration, > 5° change in inclination angle and presence of fewer than two signs of osteo-integration according to Moore criteria [16]. Heterotopic ossi cation (HO) was categorized by the classi cation system of Brooker et al [17]. The acetabular inclination and anteversion angles were measured according to the method described by Widmer [18]. 40±10° for inclination and 15±10° for anteversion were regarded as the safe zone proposed by Lewinnek et al [19]. Preoperative, postoperative, and the change in leg length discrepancy (LLD) were recorded.
The LLD is measured from the inter-teardrop line to the midpoint of the lesser trochanter on both sides.
The vertical and horizontal locations of cups were documented according to the method described by Martel et al [20]. Patients who had undergone bilateral THA or who had abnormal contralateral hips were excluded. We calculated the difference between each horizontal and vertical distance in the reconstructed hip and in the contralateral hip. All radiographs reviewed were at the latest follow-up examination from the electronic picture archiving and communication system (PACS) in our institute and the magni cation ratio was determined from the known diameter of implanted prosthetic head. Each parameter was measured twice with Mimics software (version 16.0) and averaged.
The statistical analysis was used by SPSS software (version 17.0). Descriptive analyses for categorical variables were based on percentages and frequencies and for continuous variables on mean and standard deviation (SD) or median and quartile (25-75%) if the data were skewed. The preoperative and nal follow-up HHS and WOMAC were compared using the Wilcoxon signed rank test. The correlations between continuous variables and ordinal variables were determined using Pearson correlation analysis and Spearman rank correlation analysis by correlation coe cient (r), respectively.
The method for intergroup comparisons were as follows. 101 hips with unilateral hip replacement were classi ed into an inferior migration subgroup (inferior migration distance of the reconstructed COR >5mm), superior migration subgroup (superior migration distance>5mm) and subgroup with anatomical restoration (inferosuperior migration distance equal or less than 5mm). Similarly, these hips were classi ed into a medial migration subgroup (medial migration distance >5mm), lateral migration subgroup (lateral migration distance>5mm) and subgroup with anatomical restoration (mediolateral migration distance equal or less than 5mm). For the cases with superior migration of reconstructed COR, hips were classi ed into a subgroup with anatomical restoration (0-5mm), a subgroup with mild migration (5-10mm), a subgroup with moderate migration (10-20 mm) and a subgroup with severe migration (≥10-20 mm). The statistical analysis was used to examine the differences between the variables describing these subgroups, including patient demographics (gender, age at THA, BMI, treatment interval, preoperative HHS and WOMAC), trauma-related factors (fracture pattern, initial ORIF treatment, preoperative LLD) and surgery-related factors (acetabular inclination and anteversion angle, use of an elevated liner, cup diameter, and bulk autografts). We also analyzed the differences between these subgroups regarding postoperative outcomes including LLD, Trendelenburg sign, HHS and WOMAC, and complication rate. The intergroup differences were compared by using Kruskal-Wallis After THA, the Thomas and Trendelenburg signs were positive in 8 (7.5%) and 18 hips (16.8%), respectively. All but seven patients were very satis ed (75 patients) or satis ed (22 patients) with the results. Four patients who were very dissatis ed all developed periprosthetic infection. Three patients were somewhat dissatis ed. One of them had an iatrogenic sciatic nerve injury, one patient had perceived LLD and another one endorsed persistent thigh pain.

Radiographic results
The 101 hips with unilateral THA were classi ed into a superior migration subgroup (46 hips), inferior migration subgroup (6 hips) and normal subgroup (49 hips). The results of intergroup comparisons among these three subgroups were listed in Table 3. Similarly, these hips were classi ed into a medial migration subgroup (31 hips), lateral migration subgroup (55 hips) and subgroup with anatomical restoration of COR (15 hips). The results of intergroup comparisons were listed in Table 4. In 79 hips with superior migration of reconstructed COR, the results of intergroup comparisons among anatomical restoration subgroup (33 hips), mild migration subgroup (22 hips), moderate migration subgroup (17 hips) and severe migration subgroup (7 hips) were listed in Table 5.

Discussion
Although THA after failed treatment of an acetabular fracture is a technically demanding procedure, reconstruction with cementless components can achieve satisfactory prosthesis survivorship and clinical results [8, 9,10,11,12,13]. Our clinical ndings in this young, active patient population affected by high-energy trauma were encouraging with a mean HHS of 93 and WOMAC of 5.8 at the latest follow-up examination.
A series of published studies have discussed the signi cances of the precise location of the COR in THA performed for patients with developmental dysplasia of the hip (DDH) [21], rheumatoid arthritis (RA) [22] and in revision situations [23]. However, in the published studies focusing on cementless acetabular reconstruction for failed treatment of acetabular fractures, only a few discussed the association of reconstruction accuracy in terms of hip COR with clinical results. Ranawat et al [10] reported only in three hips (9%) that the position of the COR was greater than 20 mm beyond vertical symmetry, horizontal symmetry, or both when compared with the unaffected hip. They found that nonanatomic restoration of COR was ultimately associated with need for revision surgery and with intraoperative discovery of a segmental acetabular defect. However, this did not correlate with prior treatment and HHS or with polyethylene wear.
In the current series, we found that a wide range of distribution of the reconstructed COR. The reconstructed COR was within 5 mm of vertical and horizontal symmetry (anatomical restoration of COR) in 24 hips (24%) and was more than 20 mm beyond vertical symmetry, horizontal symmetry, or both (nonanatomical restoration of COR) in 8 hips (8%). This distribution model re ected the abnormal anatomy of the acetabulum after initial fractures and treatments, and was further affected by the hardware and bone defect during implantation of the cementless cups. Consequently, compared with THA reconstruction for DDH or RA, the implant reconstruction for acetabular fractures become more irregular and technically demanding, which is analogous with situations in revision THA.
The results of horizontal intergroup comparisons suggested that only some preoperative parameters, including treatment interval, grade of acetabular bone defect and diameter of cup, had signi cant differences. For vertical intergroup comparisons, only percentage of postoperative positive Trendelenburg sign was signi cantly higher in superior migration subgroup than that in subgroup with anatomical restoration of COR (28.3% vs 8.2%, P=0.015). Moreover, the distance of superior migration of reconstructed COR was associated with positive Trendelenburg sign at the latest follow-up (r=0.504; p < 0.001). We made further investigations for 79 hips with superior migration of reconstructed COR. The percentage of postoperative positive Trendelenburg sign were equally higher in severe migration subgroup than that in anatomical restoration subgroup and in mild migration subgroup. These results suggested that a superiorly migrated COR appeared to exert a negative effect on abductor muscle function by decreasing the abductor moment arm, but did not detect any signi cant correlations with worse clinical outcomes and higher complication rate. Consequently, the abductor muscle function should be emphasized in postoperative rehabilitation program for this special patient population.
The main strengths of our study included a relatively large sample size and a high follow-up rate. However, we acknowledge that our study was not without its limitations. First, as this was a retrospective study, there are associated inherent limitations, most importantly the absence of control cases with which to compare outcomes. Patient selection, measurement and interviewer bias may affect functional assessment outcomes.
Second, although the cases enrolled in the study were exclusively from one single center, all THAs were performed by several senior surgeons, which may affect the validity of our ndings. Third, the multiple cementless acetabular component designs and different interface selections were included in our series.

Conclusions
In conclusion, our study revealed that despite the technically demanding nature and a relatively high complication rate, cementless THA performed in patients with failed treatments for acetabular fractures achieved predictable clinical and radiographic outcomes. A superiorly migrated COR appeared to exert negative