Can EOS and lateral pelvic radiographs reliably identify patient with reduced pelvic roll back at risk for dislocation ?

Abstract

Introduction

Hip instability and dislocation post-THA is a prevalent complication leading to revision surgeries. Patients are considered “high-risk” if the sit-to-stand sacral slope change is less than 10 degrees. The current study analyses if lateral radiographs and EOS imaging result in similar measurements and identify the same at-risk patients.

Materials and methods

In a retrospective study consisting of 59 hips (58 patients), spinopelvic measurements were obtained from lateral sitting and standing pelvic radiographs and EOS sitting and standing (whole body) films taken during the same preoperative assessment. Measurements on each images set were performed to assess variation in measurements between the two modalities.

Results

The average measured sacral slope difference between the two imaging methods was 4.5° (standing) and 6.6° (sitting), with a sit-to-stand difference of 7.2°. Lateral radiographs identified 16 patients at high-risk for dislocation and EOS identified 15 high risk patients; however, only 9 patients overlapped between the two modalities. There was a significant postural difference between the two modalities: average spinal angle difference was 10.4°(p= 0.0026).and femoral angle off the horizonal was 7.4°(p= 0.16).

Conclusion

There are discrepancies between EOS and conventional lateral radiographs in measuring sacral slope and change in sacral slope from sitting to standing and both imaging modalities might identify different high-risk patients for dislocation. Efforts need to be made to standardize the patients’ position at the time of imaging.

Introduction

Hip dislocation following primary total hip arthroplasty (THA) is a main reason for revision surgery [1]. In fact, the American Joint Replacement Registry’s 2024 Annual Report revealed instability was the third most common (17.4%) overall and second most common (21.1%) early indication for revision surgery following a primary THA [2]. Lewinnek et al. identified “safe zones” for acetabular component positioning and suggested that positioning the acetabular component within 40° ± 10° inclination and 15° ± 10° of anteversion would minimize the risk of postoperative dislocations [3]. However, recent studies have found that patients with components in the “safe zones” are still at risk for dislocation [4–6]. In a retrospective study by Abdel et al. consisting of 9784 patients with 206 postoperative dislocations, 58% of the dislocations had components that were in the Lewinnek “safe zones” [5].

As a result, studies have explored the relationship between sagittal spinopelvic kinematics and the risk of dislocation. During the position change from standing to sitting the physiologic rollback of the pelvis protects the hip from posterior dislocation by increasing the functional anteversion [7–9]. A decreased change of the sacral slope (SS) or the anterior plane pelvic tilt (APPt) from standing to sitting is used to identify patients at risk for posterior hip instability [7].

The Hip-Spine Classification system was created to identify patients at “high risk” for dislocation using spinal alignment, anterior pelvic plane, and spinal mobility as grouping criteria [8]. Measuring the change in sacral slope from a standing to sitting position can help identify the spinopelvic mobility as the sacral slope changes dynamically with pelvic rollback [9–11]. Normal spinal mobility is defined as at least a 10° change in sacral slope from standing to sitting whereas a “stiff spine” is defined as < 10° change in sacral slope from standing to sitting [8, 12].

Pelvic tilt significantly influences the functional orientation of the acetabular component in total hip arthroplasty [13]. Robotic systems often rely on pelvic tilt as a crucial input variable to provide alignment targets [14]. In addition, changes in pelvic position between standing and supine influence their cup anteversion and inclination targets [15]. Manual surgeons also use these variables to identify patients at risk for dislocation, which might impact their implant selection (Dual mobility).

Few studies have compared pre-operative Hip-Spine parameters on EOS imaging to those measured on standard lateral radiographs to evaluate the reliability of EOS images for preoperative planning. Lazennec et al. found that there was a difference in means in the post-THA assessment of these two imaging modalities for spinopelvic parameters, including sacral slope (SS), anterior pelvic plane tilt (APPT) in standing and sitting position and pelvic incidence (PI), of 0.30° to 3.43° (P < .05) [16]. In contrast, Shakeri et al. showed that EOS and lateral radiograph measurements of spinopelvic parameters in the standing position are comparable [18].

This retrospective study aims to determine whether (1) the spinopelvic measurements in the standing and seated position on conventional lateral radiographs are comparable to measurements on EOS imaging. In addition, the study aims to assess whether (2) both modalities will identify the same patients with less than 10 deg. of rollback and (3) report the factors that may contribute to the differences in spinopelvic measurements. We hypothesized that pelvic tilt measurements and standing-to-sitting changes obtained from EOS imaging and conventional lateral radiographs would demonstrate minimal variation, such that both modalities would identify the same patients at risk for dislocation following total hip arthroplasty.

Materials and methods

The study received Institutional Review Board approval at the authors’ institution (IRB #2024 − 1041). This is a single surgeon study, and the senior surgeon uses lateral pelvic imaging in a standing and sitting position routinely for preoperative planning of all his primary total hip arthroplasties. The study included patients examined between October 2023 and June 2024. Due to limited availability of EOS 70 patients had both EOS and lateral pelvis imaging for preoperative planning while the majority of the patients only had one of the two imaging modalities and therefore were not included in the current study. All patients underwent primary total hip arthroplasty. Selection for anterior or posterior approach are made based on body habitus and the anterior approach is avoided for patients with considerable pannus overhanging the incision area. Some patients had degenerative spine disease, and the random selection of patients for both imaging modalities due to office location resulted in a standard cohort of arthroplasty patients. 12 patients were excluded since they did not have both lateral and EOS images in the sitting and standing positions leaving 58 patients that were included in the current study. Within this group, 43 patients underwent anterior approach THA (14 Left, 25 Right, 4 Bilateral) and 15 patients underwent posterior approach THA (8 Left, 7 Right, 0 Bilateral). Three of the four bilateral patients were scheduled for a single-stage bilateral procedure. Given that the patients undergoing single-staged bilateral THAs only received one preoperative imaging study, they were counted as one set of measurements in our dataset. Patients undergoing a two-stage bilateral procedure, where each side was staged about three months apart, received two separate preoperative imaging assessments at different time points and thus were included as two distinct data points. Thus 58 patients with 59 preoperative studies were included in this study.

The study was comprised of 32 (55.9%) female and 26 (44.1%) males. The mean age at the time of the imaging studies was 62.0 years (range 32–82, SD 9.1 years) with a mean BMI of 29.0 kg/m² (range 19.1–47.4, SD 6.0 kg/m²). A chart review was performed to obtain the pre-operative physical exam findings including hip flexion and the presence of a hip flexion contracture.

Radiographic evaluation

All radiographs were taken during a single preoperative assessment. The sitting radiographs, both EOS and conventional lateral, were taken with patients seated in a height-adjustable stool. The patients were put in standard sitting positioning with their femurs parallel to the floor and 90° of flexion in both knees and hip joint (Fig. 1) [17].

Fig. 1.

Standard positioning for sitting EOS (left) and sitting lateral (right) radiograph

All radiographic measurements were completed using the orthotool-box, Sectra PACS software package. IDS7. Measurements were performed by an orthopedic resident and medical student under the guidance of a staff radiologist and the senior orthopedic surgeons. All cases with variation between both techniques were individually reviewed by the senior surgeon. Interobserver and intraobserver reliability were assessed using intraclass correlation coefficients to evaluate agreement between measurements obtained by different observers and by the same observer on repeat assessment.

Sacral slope (SS) was measured as the angle formed by the tangent line to the superior endplate of S1 and the horizontal plane [19] (Fig. 2). This was measured for all sitting and standing EOS and conventional lateral radiographs. The anterior pelvic plane tilt (APPT) was measured using the angle between the anterior superior iliac spines (ASIS) and the pubic tubercle on standing films [20] (Fig. 2). When the two ASIS were not aligned, the distance between the ASIS was measured and APPT was calculated as the angle from the pubic tubercle to the midpoint of the two ASIS [7, 20, 21] (Fig. 2). Pelvic incidence (PI) was determined by calculating the angle between the line drawn from the center between the femoral heads to the center of the S1 endplate and the line perpendicular to the S1 endplate [7]. The PI and APPT measurements were performed on standing films only. Two additional angles were measured to assess the overall sitting posture of the patient: the angle from the center of the superior S1 endplate to the center of L2, evaluating the upright posture and flexion of the pelvis, and the angle of the femur in relation to the horizontal plane, determining the degree to which the femurs deviated from standard positioning (Fig. 3). A forward tilt of the spine was defined as positive, as was a femoral angle above the horizontal plane.

Fig. 2.

Measurements for sacral slope (SS - blue) and anterior plane pelvic tilt (APPt - red)

Fig. 3.

Measurements for posture assessment

Statistics

All statistics measurements were done in Microsoft Excel. The mean values, standard deviation, minimum, and maximum were calculated for all measurements. The same calculations were also done for the difference between EOS and Lateral measurements: average difference, maximum, minimum, and standard deviation. A paired t-test was performed to determine the significance of the difference between the measurements for each modality. Additionally, a Wilcoxon Rank Sum Test (non-parametric) was performed to determine the significance of the difference in the sit-to-stand sacral slope measurements.

Results

The sacral slope measurements are displayed in Table 1. The average difference between the sacral slope measured from lateral radiographs and EOS imaging in a standing and sitting position was 4.5° (p= 0.25) and 6.6° (0.30), respectively, with a stand-to-sit sacral slope change mean difference of 7.2° (1.0) between the two modalities. The average difference between the measured APPT from each modality was 3.5° (p = 0.33) with a mean APPT on lateral films and EOS films of -1.9° and − 1.3° respectively. The mean PI was 54.1° and 56.4° respectively. There was an average difference of 5.3° (p= 0.01) in PI measurements between the two modalities (see Table 2).

Table 1.

Spinopelvic parameter measurements

		Standing			Sitting			Stand to sit
		Lateral	EOS	Lateral vs. EOS	Lateral	EOS	Lateral vs. EOS	Lateral	EOS	Lateral vs. EOS
Sacral Slope	Min (deg.)	7	10	0	-6	1	1	0	1	0
	Max (deg.)	63	66	17	55	48	27	48	47	26
	Mean (SD) (deg.)	36.5 (11.0)	37.4 (12.7)	4.5 (4.0)	19.8 (11.9)	21.0 (11.4)	6.6 (5.5)	17.1 (10.5)	16.5 (11.1)	7.2 (5.9)
	P value			0.25*			0.30*			1**
	Min (deg.)	-31	-26	0
APPT	Max (deg.)	18	19	13
	Mean (SD) (deg.)	-1.9 (8.4)	-1.3 (8.8)	3.5 (2.9)
	P value			0.33*
	Min (deg.)	33	29	0
PI	Max (deg.)	86	93	23
	Mean (SD) (deg.)	54.1 (13.4)	56.4 (14.5)	5.3 (5.1)
	P value			0.01

* Paired t test

** Wilcoxon Rank Sum test (non-parametric)

Table 2.

Assessment of the posture of the patient at the time the lateral image was taken in a sitting position

Posture
		Lateral	EOS	Lateral vs. EOS
Spine	Min (deg.)	-29	-17	0
	Max (deg.)	18	20	33
	Mean (SD) (deg.)	-4.7 (9.5)	-0.6 (8.9)	10.4 (8.6)
	P value			0.00026
Femoral	Min (deg.)	-20	-20	0
	Max (deg.)	21	8	23
	Mean (SD) (deg.)	-3.1 (8.5)	-1.9 (5.7)	7.4 (5.5)
	P value			0.16

Sit-to-stand sacral slope measurements obtained from EOS images identified 15 patients with a stiff spine (< 10° change in sacral slope from standing to sitting) at high risk for dislocation while those from lateral radiographs identified 16 patients as high-risk, with only 9 patients overlapping as high-risk between the two modalities. For patients at high-risk, the senior surgeon adjusts cup position but does not change the bearing option or approach.

Details of the patient position during sitting images are displayed in Table 2. The average difference in spinal posture between EOS and Lateral films in the sitting position was 10.4° (0°, 33°, p= 0.0026). The mean value of the femoral angle off the horizontal was 7.4° (0°, 23°, p= 0.16).

Both interobserver and intraobserver reliability were excellent. The overall interobserver intraclass correlation coefficient was 0.995, and intraobserver reliability demonstrated similarly high agreement, with intraclass correlation coefficients consistently exceeding 0.99.

Discussion

The current study shows that, while not statistically significant, there were clinically relevant differences between the two imaging modalities when measuring spinopelvic parameters. Sit-to-stand sacral slope measurements obtained from EOS and standard lateral radiographs identified different patients with a stiff spine at risk for dislocation following total hip arthroplasty.

When assessing spinopelvic measurements between EOS whole spine images (standing) and standing lateral radiographs, Wu et al. found that the average difference between the imaging studies was < 1° for all spinopelvic measurements in an analysis of 50 patients [22]. The present study extends this comparison by incorporating additional postures and a broader set of parameters. The inclusion of sitting images, introduces postural and positional variability that may account for the larger differences observed in the current cohort.

Modern robotic and navigation algorithms rely on differences in standing and sitting positions to predict impingement and adjust the acetabular anteversion. Many robotic systems consider these variables accurate and adjust cup position based on the measurements taken from radiographs; however, the current study suggests these numbers can be quite variable and might be influenced by the flexibility of the hip before and after surgery. Additonally, for patients considered high-risk, the senior surgeon may adjust cup position based on perceived instability risk, while not altering the surgical approach or bearing selection. The lack of reproducibility of measurements between the two different imaging modalities suggests that care should be taken not to overly rely on these measurements for the adjustment of acetabular positioning parameters. Given the variation in measurements reported in the current study care should be applied when adjusting acetabular anteversion and performing impingement modeling based on standing and sitting radiographs. Additionally, the inaccuracies in identifying high-risk patients need to be taken into consideration during preoperative planning [17, 23].

The current study suggests that a number of factors are responsible for differences in measurements: (1) differences in stool height can influence the amount of hip flexion, (2) the position of the upper body can influence sacral slope by reducing rollback when leaning forward or increasing rollback when leaning backwards and (3) stiffness of the hip can result in increased rollback when attempting to bring the femur in a parallel position to the floor which forces the sacrum into more rollback position then explained by spinopelvic mobility and (4) pelvic rotation on standing and lateral radiographs can make the identification of the sacral endplate or APPT more difficult. There is also inherent difficulty in controlling pelvic rotation during acquisition of lateral pelvic radiographs, which can affect visualization of the sacral endplate and anterior pelvic plane reference points.It is crucial to further investigate techniques to standardize the sitting position when doing seated lateral pelvis radiographs and identify factors to identify patients at risk for increased differences in measurements [24, 25].

The current study has the following limitations: (1) this study was not designed to determine whether spinopelvic measurements obtained from EOS imaging or conventional lateral radiographs are more accurate in predicting postoperative total hip arthroplasty dislocation. Additionally, information regarding intraoperative decision-making and component positioning adjustments was not available, limiting the ability to correlate preoperative imaging findings with surgical technique or clinical outcomes. (2) While all efforts are made to standardize patient position during the different radiographs the large number of technicians performing radiographs at the authors’ institution might result in differences in technique between technicians. (3) while small differences in intra and interobserver reliability of measurements were observed, these were not meaningful enough to explain the rather clinically meaningful differences observed in measurements in the current study.

Conclusion

The current study at a high-volume institution suggests discrepancies in measurements of sacral slope and anterior pelvic plane are observed between EOS imaging and standard lateral pelvis radiographs. Of concern is that both techniques identify different patients as high-risk for postoperative dislocation. Both techniques depend significantly on the technician adjusting the position of the patient at the time of imaging. The current study suggests that more efforts need to be made to standardize patient positioning as well as imaging techniques and to identify patients at risk for measurement variations. Surgeons need to be aware of these discrepancies when using changes in pelvis position from standing to sitting to model impingement or adjust the acetabular anteversion in total hip arthroplasty.

Author contributions

Conceptualization, F.B. and T.M. and K.S.; methodology, K.S. and F.B.; data curation: K.S. and C.M.S. and A.M., validation: F.B and T.M.; writing original draft preparation K.S.; writing-review and editing, F.B. and T.M.; supervision F.B. All authors have read the manuscript, attest to the validity and legitimacy of the data and its interpretation, and agree to its submission.

Funding

One Author (KS) received funding for conducting this study as part of the Medical Student Summer Research Fellowship at Hospital for Special Surgery.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Conflict of interest

The authors declare no competing interests.