Impact of stem design and cementation on postoperative femoral antetorsion in 227 patients with total hip arthroplasty (THA)

This retrospective study included consecutive patients that underwent MRI from February 2015 until October 2019 and was approved by the Zurich cantonal ethics committee. A total of 263 patients with an age above 18 years and with hip prostheses on either one or both sides were included. The following inclusion criteria were applied: availability of plain radiograph for the verification of stem type and hip MRI with an additional axial series over the knee for antetorsion measurements. Patients were operated on and referred to our institution either from the in-house orthopedic department or from an external partner hospital. All imaging was performed at our institution. The following exclusion criteria applied: artifact significantly limiting evaluation for anteversion. Stem loosening was defined as ≥ 2 mm of increased signal between femur and prosthesis stem or between femur and cement in the case of a cemented stem on fluid sensitive sequences. An example of suspected stem loosening is given in Fig. 1. Loosening may lead to secondary rotation of the stem in the femur and falsify the initial postoperative femoral antetorsion.

The study design is presented in Fig. 2. Five different patient groups according to five different stem types from different manufacturers with or without cementation were analyzed; stem 3 and stem 4 were from the same manufacturer but had different designs. The groups were labeled stem 1 to stem 5 (S1–S5). Stem 1 Fitmore, Zimmer Biomet Holdings Inc., Warsaw, Indiana (S1, n = 32), was cementless and short curved. Stem 2 Quadra, Medacta Group SA, Castel San Pietro, Switzerland (S2, n = 51), and stem 3 Corail, DePuy International Ltd., Leeds, England (S3, n = 53), were cementless and standard straight. Stem 4 Corail Collared, DePuy International Ltd., Leeds, England (S4, n = 48), was collared cementless standard straight and stem 5 (S5, n = 43) was cemented straight from two different manufacturers: Avenir, Zimmer Biomet Holdings Inc., Warsaw, Indiana, and Quadra, Medacta Group SA, Castel San Pietro, Switzerland. For method validation, the accuracy of femoral antetorsion measurements in MRI was tested against CT, which is the current gold standard. In 30 cases, femoral antetorsion was measured in CT and MRI in the same patients. Measurements of the femoral antetorsion in the CT were started 2 months after the MRI measurements were completed to reduce recall bias by the two readers. Femoral antetorsion was measured as described below.

All patients underwent MRI of the hip on a 1.5T Magnetom Avanto-fit system (Siemens Healthcare, Erlangen, Germany). For all hip MRI examinations, patient positioning is standardized. After placement of the patient in a supine position, a tape is carefully put around the feet of the patient in order to fix the position of the lower extremity in a slight internal rotation of the hip during the scan. The MR protocol was specially tailored for THA (Table 1) and included strictly axial T1-weighted imaging with high bandwidth for metal artifact reduction at the level of the joint. To assess femoral antetorsion, a fast T2-weighted, strictly axial haste sequence at the level of the knee was used. No manual tilting of these sequences was performed during imaging acquisition. A body matrix phased-array surface coil and the integrated spine matrix coil were used for imaging of the hip. For the knee, the integrated spine matrix coil was used.

Evaluation in respect to potential shaft loosening was done on coronal short tau inversion recovery (STIR) image with compressed sensing and slice encoding for metal artifact correction (CS-SEMAC) [13‐15], as well as an axial high-bandwidth STIR sequence with an optimized inversion pulse [16].

CT scans were performed on a Somatom Edge Plus system (Siemens Healthcare, Erlangen, Germany). Patient positioning was similar to positioning in the MRI with lower extremities in fixed, slight internal rotation. The following scanning and reconstruction parameters were used: voltage, 120 kV using 4D care dose; pitch, 0.8; slice thickness, 1.5 mm; increment 1.0 mm; FOV 200 × 200 mm² at the level of the hip and the knee (femoral condyles to joint space). For reconstruction, Kernel Br51 and Strength/Safire 3 were used.

Two fellowship-trained musculoskeletal radiologists with 6 and 7 years of experience in musculoskeletal radiology independently evaluated all patients on our institution’s Merlin PACS (Phoenix-PACS GmbH, Freiburg, Germany).

For angle measurements, the PACS toolbox was used. First, the head of the prosthesis was delineated with a circle on multiple axial sections simultaneously for defining the center of the head on the whole image stack. At the level of the prosthesis neck, the midpoint between the anterior and posterior contour served as the second reference point. Subsequently, an angle measurement was performed between the proximal and distal femur as follows: (i) proximal line through the prosthesis neck and the center of the prosthesis head and (ii) distal line, which was the posterior condylar tangent of the knee on the second imaging series [17]. An example is given in Fig. 3. CT measurements of femoral antetorsion were done similar to MRI measurements.

Statistical analysis was performed on SPSS version 21.0 (IBM Corp, Armonk, NY) and MedCalc version 17.6 (MedCalc Software bvba, Ostend, Belgium). For continuous data, general descriptive statistics were reported as means and standard deviation (SD). Normal distribution of the measured angles, patient age, and sex was evaluated with the Kolmogorov-Smirnov test; for statistically significant differences in femoral antetorsion, the t test for independent samples was applied. The Kruskal-Wallis test checked for differences in age, gender, and occurrence of retrotorsion. Postoperative torsional variability for each group is represented by the SD; we used the F-test to check the SD in the cementless and the cemented group for statistically significant differences. In all tests, a p value of < 0.05 was considered to represent statistical significance.

The two-way random-effects intraclass correlation coefficient (ICC) was applied for inter-reader agreement, whereas ICC values > 0.75 were considered good agreement, and > 0.9 as very good [18].

A total of 263 consecutive patients were reviewed for inclusion in the study. In 27 cases, metal artifacts at the level of the head of the prosthesis were very heavy and it was not possible to delineate the femoral head precisely. Inaccurate localization of the center of the femoral head would lead to inaccurate measurement of the femoral antetorsion and therefore led to exclusion. Suspected stem loosening led to the exclusion of another 9 patients, resulting in a final set of 227 cases.

Distribution of age and gender among the different groups is given in Table 2. The median age of the 227 patients was 64.0 years (range 30.0–87.0 years, interquartile range (IQR) 55.0–74.0). The overall age was not normally distributed; this variable was considered non-parametric. Using the Kruskal-Wallis test, age in the cemented group S5 was significantly higher compared with the non-cemented S2 (p = 0.001), S3 (p = 0.031), and S4 (p < 0.001).

Overall, there were 103 males and 124 females included. There were significantly more females in S5 compared with those in S1 (p < 0.001), S2 (p = 0.001), S3 (p < 0.005), and S4 (p < 0.005).

Cases from all subgroups were included as follows: stem 1, 7 cases; stem 2, 7 cases; stem 3, 6 cases; stem 4, 6 cases; and stem 5, 4 cases. Mean time between CT and MRI examination was 7.97 months. In 14 cases, CT was prior to the MRI examination; in 15 cases, MRI was prior to the MRI examination; in one case, MRI and CT were performed on the same day.

ICC was calculated between CT and MRI for reader one and reader two. For both, ICC was very good with an ICC of 0.98 (0.96; 0.98) and 0.91 (0.79; 0.96), respectively.

The ICC was very good with a value of 0.98 (0.973; 0.984), and the measured femoral antetorsion angles of both readers were averaged. Angles were normally distributed; this variable was considered parametric.

Results, standard deviation, and range are shown in Table 3 and Fig. 4. Antetorsion was highest in the collared, cementless standard straight group (S4) with a mean value of 18.1° and lowest in the cemented straight stem group (S5) with a mean value of 12.7°. The measured angles were significantly different between S4 and S5 (p = 0.007). The cementless short curved (S1) and both cementless standard straight (S2 and S3) were below S4 and above S5, with mean values of 14.8°, 14.1°, and 13.3° with no statistically significant difference. A significant difference was found between S3 and S4 (p = 0.012).

In all prosthesis types, the range and the SD were high. Figure 5 shows the distribution of the femoral antetorsion for the different stems. For S1, 9 of 32 (28.1%) stems were positioned in the target range of 10° to 20° during surgery; S2, 15 of 51 (29.4%); S3, 19 of 53 (35.8%), S4, 20 of 48 (41.7%); and S5, 19 of 43 (38.8%). Except for S1, positioning in retrotorsion occurred: S2, 6 of 51 (11.8%); S3, 2 of 53 (3.8%); S4, 2 of 48 (4.2%); and S5, 1 of 43 (2.3%). Positioning in retrotorsion was statistically different among the different groups (p < 0.001). A comparison of the combined SD between all cementless stems (S1–S4) and the cemented group (S5) was significantly different (p = 0.019). SD was highest in one of both cementless standard straight (S2, SD = 12.2) and lowest in the cemented group (S5, SD = 7.7), also with a significant difference (p = 0.006). Significant differences in SD were seen for both standard straight stems (S2 and S3) (p = 0.018).

This study confirmed that there is a wide variety in postoperative femoral torsion in patients with THA.

The measured antetorsion in the short curved (S1, 14.8° ± 10.0°) and both collarless standard straight stems (S2, 14.1° ± 12.2° and S3, 13.3° ± 8.4°) were only slightly elevated compared with the femoral antetorsion in healthy individuals, described by Sutter et al. [3] with similarities in SD (12.7°–13.5° ± 9.8°–10.8°). Antetorsion values in the collared stem group (S4, 18.1° ± 10.5°) were explicitly higher compared with all other groups with statistical significance to the cemented stems (S5) and to one of both cementless collarless standard straight stems (S3). Remarkably, S3 and S4 are of the same manufacturer, only differing in the presence or absence of a collared neck. When introduced into the femur, both stem types (S3 and S4) are likely to have equal physical properties and behave in a similar way, the collar is unlikely to directly influence antetorsion during the introduction process, and placement in elevated antetorsion in S4 cannot be attributed to stem design below the collar. Still, there is a significant difference in femoral antetorsion between both groups. Before the introduction process of the stem into the femur begins, the surgeon visually assesses the antetorsion angle and targets an angle of 15° antetorsion. Intraoperative estimation of stem torsion on a visual basis alone is known to be difficult even for experienced surgeons. In a recent study intraoperative, visual assessment led to the overestimation of femoral antetorsion. Estimated stem torsion was an average of 7.3° higher than CT measurement [31]. Based on our data, we suspect that during surgery the presence of the collar might impair the surgeon’s intraoperative angle estimation and leads to intraoperative visual miscalculation which results in increased postoperative antetorsion.

In S2, there was the highest frequency of postoperative retrotorsion (11.8%) and the highest absolute retrotorsion (− 20.2°), whereas no retrotorsion could be observed for the short curved stem type (S1). During surgery, when the stem is introduced into the femur, torsion towards misalignment starts after the stem has locked and is further introduced. Exceptional malpositioning in retrotorsion might be less likely in a short curved stem because the short stem is put less deep into the femur compared with a long shaft.

For orthopedic surgeons, knowledge of stem specific properties may help to achieve a better surgical outcome. Based on our data, the collared standard straight stem (S4) is more likely to be placed in elevated antetorsion compared with other groups. Future studies may be able to demonstrate whether the insertion procedure for the S4 stem can be modified to achieve a more physiologic femoral antetorsion.

In the group with the cemented stem (S5), the postoperative femoral antetorsion with a mean of 12.7° is similar to the mean femoral antetorsion in healthy individuals described previously [3]. The standard deviation of femoral antetorsion in cemented stems was lowest and was statistically significantly lower compared with the combined standard deviation of all cementless stems. Positioning in retrotorsion was uncommon (2.3%). A likely explanation is that the cement—while still liquid—allows a certain degree of rotation during and even after complete stem placement which results in a more precise placement. This finding is consistent with the hands-on experience of the surgeons at our institution.

Due to the better control over postoperative femoral antetorsion, we believe that the intraoperatively targeted angle was reached to a higher degree and might be a closer approximation to an ideal postoperative situation compared with the other groups. Statistically significant differences were observed between S4 and S5; we think this is attributed to the rather low standard deviation in the cemented group, which made this difference statistically significant.

In our study, patients with a cemented prosthesis were older compared with the cementless prosthesis: This is a finding in accordance with the clinical indications of hip replacement [32]. In this study, design patient age is most likely not a confounding factor.

This study has the following limitations: Preoperative femoral antetorsion was not known in the included patients, due to the retrospective study design. Although the included patients were all operated by specialized hip surgeons, there was no subgroup analysis for individual surgeons. In group 5, the cemented standard straight design included stems from different manufacturers due to the limited availability of cases from one single manufacturer. There were not an equal number of patients in each group; especially in group 1, there were fewer patients compared with groups 2–5. Still, in this group, the SD lies within the range of the other groups and a statistical effect might be negligible. Method validation through antetorsion measurement in MRI and CT was done in a subset of patients.