Clinical Impact of Pelvic Malrotation on X-ray based Preoperative Planning for Total Hip Arthroplasty: A Proof-of-concept and Prudent Prediction of Acceptable Rotation


 IntroductionPelvic rotation (PR) on preoperative X-ray templating can affect various critical measured acetabular angles and potentially outcomes of successful total hip arthroplasty (THA). Optimising anatomical reconstruction of the joint is essential to achieve function, longevity and prevention of complications following surgery. There is limited literature that standardises the degree of acceptable PR on X-ray or its effects on the fitting of acetabular prostheses. ObjectiveThis study aimed to develop a proof-of-concept that quantifies how PR can affect various acetabular angles used in pre-operative THA templating and to formulate a practicable method of determining if the preoperative PR is acceptable. Materials and MethodsComputerised tomography (CT) models from three control and two THA patients were generated and manipulated in various degrees of PR. CT slices were thickened to simulate X-rays and acetabular angles measured. ResultsThe acetabular anteversion distance (AAD) and lateral opening angle (LOA) demonstrated a linear and quadratic relationship with good correlation (R2=0.923, R2=0.710 respectively, p<0.0001) in relation to PR. Change in area of prosthesis (AOP) demonstrated a good linear correlation (r2=0.774 and r2=0.875, p<0.0001) with PR. Two novel measurements were used to estimate the degree of PR from a pelvic X-ray; the horizontal distance between pubic symphysis and middle of sacrococcygeal joint (PSSC) and the simplified pelvic rotation ratio (SPRR). A strong correlation between PSSC and SPRR with change in PR was observed (R2=0.970, R2=0.953, p<0.001). ConclusionPreliminary results suggests that an SPRR >2.0 correlates to PR>20° with potential to have a clinical impact on preoperative measurements.


Introduction
Pelvic rotation (PR) on preoperative X-ray templating can affect various critical measured acetabular angles and potentially outcomes of successful total hip arthroplasty (THA). Optimising anatomical reconstruction of the joint is essential to achieve function, longevity and prevention of complications following surgery. There is limited literature that standardises the degree of acceptable PR on X-ray or its effects on the tting of acetabular prostheses.

Objective
This study aimed to develop a proof-of-concept that quanti es how PR can affect various acetabular angles used in pre-operative THA templating and to formulate a practicable method of determining if the preoperative PR is acceptable.

Materials and Methods
Computerised tomography (CT) models from three control and two THA patients were generated and manipulated in various degrees of PR. CT slices were thickened to simulate X-rays and acetabular angles measured.

Results
The acetabular anteversion distance (AAD) and lateral opening angle (LOA) demonstrated a linear and quadratic relationship with good correlation (R 2 =0.923, R 2 =0.710 respectively, p<0.0001) in relation to PR.
Change in area of prosthesis (AOP) demonstrated a good linear correlation (r 2 =0.774 and r 2 =0.875, p<0.0001) with PR. Two novel measurements were used to estimate the degree of PR from a pelvic X-ray; the horizontal distance between pubic symphysis and middle of sacrococcygeal joint (PSSC) and the simpli ed pelvic rotation ratio (SPRR). A strong correlation between PSSC and SPRR with change in PR was observed (R 2 =0.970, R 2 =0.953, p<0.001).

Conclusion
Preliminary results suggests that an SPRR >2.0 correlates to PR>20° with potential to have a clinical impact on preoperative measurements.

Background
Total hip arthroplasty (THA) is a well-established method to treat hip pain and physical limitations due to osteoarthritis and other medical conditions. The acetabular prosthesis implantation angle affects muscle strength, gait, limb lengths, impingement, noise generation, loosening, postoperative range of movement (ROM) and is related to dislocation and liner wear [7]. Optimising biomechanical and anatomical reconstruction of the joint is essential to achieve function, longevity and prevention of complications following surgery [1][2][3][4][5][6]. Preoperative templating from a plain anterior-posterior radiograph is the primary method for initial evaluation and cornerstone for prosthesis positioning, however such images are subject to patient pelvic tilt and rotation. Suboptimal acetabular component position signi cantly impacts the results of a hip arthroplasty, including increased risk of instability, impingement, dislocation and cup failure [1,3,6,[8][9][10][11][12][13]. Correct template positioning in uences the accuracy of acetabular cup placement planning and long-term success of the total hip arthroplasty [6].
Effects of pelvic position on pelvic tilt has been well studied [14][15][16]. When the pelvis tilts posteriorly, the anteversion and abduction angles of the acetabular implant increases, potentially leading to excessive wear due to neck impingement and edge loading, with increased risk of anterior dislocation [6]. Prediction of pelvic displacement before surgery is therefore essential for accurate placement of the acetabular implant. In contrast, there is a lack of literature describing the effects of X-ray pelvic rotation (PR) on preoperative acetabular angulation of acetabular prostheses. The purpose of this study was to generate 3D CT pelvis models from healthy controls and arthroplasty patients and to quantify the potential effects of pelvic rotation on acetabular cup position in various planes. Furthermore, this study aimed to determine if there was a simple and practicable method of determining pelvic rotation on pre-operative templating. This proof-of-concept study hypothesised that PR may have detrimental impacts on acetabular angles during preoperative planning for THA. It also hypothesised that actual PR can be estimated from recreated plain X-rays using simple measurements and anatomical landmarks.

Study Design:
This was a retrospective case-control experimental design to investigate the effect of PR on acetabular angulation.

Participants and setting:
To establish a proof of concept, a convenience sample of control and patients undergoing total hip arthroplasty (THA) due to osteoarthritis from the Royal Adelaide Hospital was selected between the 1st January 2016 and 31st June 2018. In total ve patient's pelvic CT images were analysed, with three controls and two THA patients. The control patient CT results were con rmed normal with no abnormalities by consultant radiologist prior to inclusion in the study. The THA CT images were also con rmed to be otherwise normal prior to inclusion. Baseline patient characteristics are recorded in Table   1. De nitions used to describe the angles and measurements used in this study can be found in Table 2.
Assessment of pelvic rotation using computerised model: True pelvic rotation was established using Carestream Picture Archiving and Communications Systems (PACS) and Radiology Information System (RIS) version 11. 0.14.35. For the acquisition of the images used in this study, thickened CT slices were used to simulate the X-ray radiographs in a 2-dimensional plane. The CT reconstruction was manipulated at varying degrees of rotation from the neutral position. The neutral PR was de ned by ischiotuberosities being parallel to reference [19]. All pelvic images were aligned with a neutral pelvic tilt as described by Marratt et al. (2015) [18]. All parameters were measured on left hips. Pelvic rotation at increments of 10° was performed up to 60° in order to optimise interpretation data. Rotation was de ned as positive (+) rotation corresponding to pelvic rotation to the left (relative pelvic External Rotation, ER) and negative (-) rotation corresponding to pelvic rotation to the right (Internal rotation, IR). At each 10° increment, the AP and lateral 2-dimensional XR images of the pelvis was recorded. Respective angles, measurements and ratio calculations were undertaken using the de nitions summarised in Table 2.

Pelvic rotation (PR):
PR on AP X-ray was estimated using the horizontal distance between the pubic symphysis and middle of the sacrococcygeal joint (PSSC) using the methods described in Tannast et al. (2005) [20]. We proposed a re ned method for PR assessment using the ratio between the PSSC and the horizontal distance between the two anterior superior iliac spines (HASIS), thus allowing a more patient speci c determination. We named this measurement the simpli ed pelvic rotation ratio (SPRR).

Statistical analysis:
Statistical analysis was performed with the Statistical Package for the Social Sciences (SPSS) for Windows, Version 24.0, IBM Corp. Released 2016. Armonk, NY. A highly precise value of p <0.001 was accepted as indicating statistical signi cance. Data was expressed as mean ± standard deviation for continuous variables and as a number and percentage for categorical variables. Categorical data between the groups was compared using the Chi-squared test or Fisher's exact test, while continuously distributed data was compared using either a Mann-Whitney U test or an analysis of variance (ANOVA). Linear and quadratic trend lines were generated using SPSS and reported using a R-squared value. Line goodness of t was assessed using an ANOVA linear or quadratic regression and reported using an Fvalue with a signi cance level of p<0.001 used.

Page 5/15
Acetabular measurements: There were no statistically signi cant differences in the baseline characteristics of the control hips and THA groups used to generate CT models (Table 1).  The acetabular anteversion distance (AAD) and lateral opening angle (LOA) demonstrated a linear and quadratic relationship with good correlation (R2=0.923, R2=0.710 respectively) in relation to pelvic rotation ( Figure A, Graphs 1 and 3).

Discussion
The AP radiograph remains the mainstay initial evaluation of pelvic anatomy for preoperative THA planning. Perceived pelvic anatomy and associated angles and measurements on an AP radiograph depend directly on the extent of PR. This study aimed to determine the relationship between PR and such measurements, if there is there an acceptable degree of PR, and if there is there a simple method of quantifying PR from an AP pelvic radiograph. Using the CT reconstructed images, the degree of pelvic rotation was able to be manipulated and recorded with high precision. By neutralising pelvic tilt completely, the potential confounding effect of pelvic tilt on acetabular angles are ameliorated.
For the purposes of this discussion in the context of interpretation in preoperative THA planning, trendlines of Graphs 1-6 ( Figure 1) where analysed and approximations were made between the angles of 0 to 40° relative pelvic internal rotation (IR) and external rotation (ER). We recognise that there are number of different methods to assess acetabular anteversion in native and THA patients [21]. Our data suggests that the observed AAD (Graph 1) increases linearly by approximately 2-3° for every 4° of pelvic rotation in either direction. It is clinically necessary to ensure an appropriately neutral pelvis to avoid falsely interpreting excessive anteversion. We suggest that up to 20° of PR will lead to a safe estimation of AAD.
In contrast, the lateral opening angle (LOA) tends to increase linearly from relative pelvic IR to ER by approximately 1° for every 4° of PR. Rojas et al. (2018) [22] found that an altered pre-operative Sharp's angle (methodologically similar to LOA [23] ) was associated with cup malpositioning (OR 2.51, p=0.02). It is therefore imperative to ensure minimal PR and therefore accurate preoperative LOA to avoid potential cup malpositioning and therefore potentially poor THA outcomes. There is a sparsity of literature reporting the effect of PR on LOA, however our results are consistent with one study's conclusion that PR increases the error of acetabular index in 3-month olds with developmental dysplasia of the hip [24]. A study from Tannast et al. (2015) [19] that applied a similar method using CT imaging demonstrated methodological accuracy with good reproducibility and reliability. It also revealed Sharp's angle did not have a clinically signi cant change between neutral to 12° (bilaterally), however their experiment did not report beyond 12° of PR. Clinically we suggest that up to 20° of PR will lead to minimal impact on the nal measurement of LOA.
The AOP tends to increase linearly from relative pelvic ER to IR by approximately 1cm 2 for every 4° of PR. The AOP for the two different implant sizes in in Graph 4 must be analysed separately in order to be clinically meaningful, yielding good linear correlation with PR. This has the potential clinical implication that a relative pelvic internal rotation can result in an under estimation of post-operative acetabular prosthetic coverage and therefore implant size choice. Likewise, PR of less than 20° is unlikely to pose a signi cant negative impact on AOP estimation. It is also recommended that future analyses group datapoints for similar implant sizes to reduce unnecessary variation.
Taking into consideration that both AIA and AD demonstrated a weak correlation with PR, the AIA tends to increase linearly from relative pelvic ER to IR by approximately 1° for every 5° of PR. AD tended to follow a quadratic trendline, where both IR and ER of the pelvis reduced observed AD by approximately 1.5mm per 10° of rotation. A study by Ghostine et al. (2016) [25] investigating 3-D reconstructions from biplanarradiographs also reported that AIA and acetabular anteversion was minimally affected by PR from neutral to 20° in comparison with other hip parameters, thereby supporting our results. The clinical applicability of our AIA and AD data is however tenuous.
As previously discussed, incorrect acetabular prosthetic cup positioning is known to be associated with poor clinical outcomes, such as impingement, instability or prosthetic failure [1,4,2,3,5]. It is therefore important to determine an acceptable degree to PR that will not signi cantly bias preoperative planning. Synthesising results from this proof-of-concept study, PR of more than 20° can produce unreliable acetabular angles and measurements, leading to increased risk of incorrect prosthesis placement. It is therefore imperative to ensure a safe amount of PR on AP radiograph.

Estimation of rotation:
In this study, actual pelvic rotation determined by CT can be estimated on AP X-ray using the horizontal distance between the pubic symphysis and middle of the sacrococcygeal joint (PSSC) and the horizontal distance between the two anterior superior iliac spines (HASIS). The results show a near-perfect correlation between PSSC and PR, giving a clear indication to the extent of PR on an AP radiograph. These results are comparable to a previous study [20] that employed similar methodology on 20 cadavers. Using the trendline from Graph 5, a PSSC of less than 10mm is expected at 0° of pelvic rotation, with an increase of approximately 20mm for every 10° of PR in either direction.
It can be argued however that these measurements are subject to individual patient anthropometry, gender, size, radiographic technique and analysis software, therefore a more patient-speci c method was investigated. By using a novel ratio, the SPRR, patient factors and imaging techniques were expected to have less overall impact. Using Graph 6, at neutral pelvic rotation the SPRR is shown to be less than 1. As PR increases in either direction, PSSC (measured in mm) increases and HASIS (measured in cm) decreases, therefore the ratio increases. Analysing preliminary data used in this study, an SPRR greater than 2.0 is therefore likely associated with X-ray pelvic rotation of more than 20°. It would therefore seem appropriate for this value to be employed as a safe cut off for PR tolerance. Limitations: This novel proof-of-concept study does have some methodological limitation, including a limited sample size of pelvises, single observer and unilateral hip measurements. It would therefore be questionable to generalise the exact values to a particular age bracket, gender or speci c diagnostic indication for THA. The 3-D pelvis models generated we used was based on real human pelvises, therefore are subject to individual anatomical differences. The goal of this study however was not to necessarily to report the absolute acetabular angles, but rather to demonstrate the effect of PR on these values. We reasoned that pelvises of differing anatomical structure would have similar degrees of changes due to PR, regardless of the absolute acetabular angle measurement. Despite these limitations therefore, the trendlines discussed demonstrated acceptable goodness of t and clinically useful data for, AAD, LOA, estimation of rotation using PSSC and SPRR. Based on these outcomes we were able to provide an estimate the potential acceptableness of PR from an AP pelvis radiograph.

Conclusion
Pelvic rotation can signi cantly impact on the perceived acetabular angles observed on an AP pelvic radiograph, which can in turn result in poor prosthetic placement and potentially subsequent poorer longterm clinical outcomes. Data from our initial study indicates that PR of less than 20° is unlikely to have a clinical impact of preoperative measurements and therefore serve as a guide for clinical application.
We propose that our new and novel ratio of SPRR of more than 2.0 may serve as a simple and reproducible surrogate for measured PR of >20°. Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
Not applicable Funding Not applicable