Standing Anterior Pelvic Tilt is Correlated with the Proximal Femur Shape of Individuals with Cam Morphology

Abstract

Background

Spinopelvic morphology and alignment have been theorized to play a role in causing femoroacetabular impingement syndrome symptoms in individuals with cam morphology. However, prior studies have used hip morphological and alignment measurements that are interdependent, limiting understanding of form-function relationships.

Methods

Statistical shape modeling and biplane videoradiography were used for accurate, 3D assessment of femur and pelvis morphology and in vivo standing hip and pelvis orientation among individuals with and without cam morphology and impingement symptoms. Group differences in standing posture and shape were assessed, along with correlations between shape and posture.

Findings

Patients stood with less anterior pelvic tilt and greater hip extension than asymptomatic individuals without cam morphology. A more patient-like proximal femur shape was correlated with less anterior pelvic tilt. Asymptomatic individuals with cam morphology showed no standing posture differences but had a unique pelvic shape featuring an internally rotated ilium. Further, femur and pelvis anatomical features, primarily describing changes at hip-crossing muscle attachment sites and the sacroiliac region, showed significant associations with hip and pelvis orientation in the sagittal and frontal planes.

Interpretation

Patients with symptomatic impingement due to cam morphology may preemptively adopt less anterior pelvic tilt to enable functional hip range of motion. The associations between shape and standing posture may identify patients best suited for treatment involving pelvic posture modification. Results support further investigation of hip crossing muscle characteristics and spinopelvic alignment during dynamic activities to strengthen understanding of the underlying mechanisms responsible for symptoms in patients with femoroacetabular impingement syndrome.

Introduction

Malalignment and dysfunction of the spinopelvic complex contribute to several orthopaedic disorders [1,2] that negatively impact quality of life [3,4]. Research has shown that spinopelvic function and morphology are associated with femoral asphericity, or cam morphology [5-12], which is a recognized cause of hip osteoarthritis (OA) present in patients diagnosed with femoroacetabular impingement syndrome (FAIS) [13-15]. Conventional diagnosis and surgical treatment of cam FAIS aims to resolve painful impingement through removal of the cam lesion, without specific consideration of the spinopelvic complex. Identifying relationships between cam morphology and spinopelvic alignment could clarify the pathophysiology of cam FAIS, a mechanism obscured by prevalence of cam morphology in individuals without FAIS symptoms [16,17]. Including asymptomatic individuals with cam morphology in analyses relating hip shape to spinopelvic orientation may help to distinguish specific form-function relationships underlying symptom-inducing versus symptom-avoidant compensation strategies. Clarifying these mechanisms may support targets for conservative intervention and lays the groundwork for phenotyping FAIS symptomatology to enable more patient-specific treatment strategies.

Several studies have investigated hip, pelvis, and spine morphology and alignment in patients with cam FAIS using plain-film radiographs [7-9,18,19] and skin-marker motion capture [10,12,20]. However, radiographic measurements often fail to capture the full 3D complexity of skeletal deformities [21,22] and pelvic orientation and suffer from poor repeatability [23,24]. Additionally, conventional measurements of pelvic morphology and orientation are interdependent [9,25-28], which can confound understanding of form-function relationships. While skin-marker motion capture enables 3D quantification of hip and pelvic orientation, the calculated hip joint angles are influenced by inaccurate estimation of the hip joint center [29] and soft tissue artifact [30], which lead to erroneous measurements of hip kinematics [31]. Additionally, skin-marker motion capture typically uses generic bone morphology to visualize and measure kinematics, which limits understanding of the relationship between form and function.

Collectively, there is a need for objective analyses of femur and pelvis 3D morphology together with accurate measurements of in vivo hip and spinopelvic alignment. Greater morphological understanding can be achieved with statistical shape modeling (SSM), which has been previously demonstrated to objectively quantify 3D anatomic variation within a population [32-34]. SSM can be used to identify regional anatomical differences between groups and highlight the contributions of each group to specific morphological features. Regarding in vivo hip and spinopelvic alignment, model-based tracking of biplane videoradiography (BVR) offers an accurate [35] and reliable approach that, unlike skin-marker motion capture, is unaffected by soft tissue artifact. Importantly, bone orientation quantified by BVR is displayed relative to patient-specific bone surface reconstructions, enabling direct assessments of form-function relationships. Prior studies have utilized BVR to better understand the pathological kinematics in a cam FAIS population [5,36-38]; combining BVR with objective morphological assessment from SSM would strengthen the understanding of form-function relationships in the hip.

The objective of this study was to apply SSM and BVR in concert to evaluate how femur and pelvis morphology are associated with hip and pelvic standing posture in patients with cam FAIS, individuals with asymptomatic cam morphology (i.e., positive controls), and asymptomatic individuals without cam morphology (i.e., negative controls). To achieve our objective, we tested the null hypothesis that there were no significant differences in standing posture and no associations between SSM metrics and hip or pelvic alignment while standing.

1. Methods

2.1. Participant Population and Screening

Thirty-two individuals were recruited as part of an institutional review board-approved study (University of Utah, Salt Lake City, UT, USA. IRB #51053). Participants included cam FAIS patients (n=15), positive controls (n=7), and negative controls (n=10). All negative controls and six patients were initially recruited as part of previously published studies [5,39], while all positive controls and nine patients were newly recruited (Table 1). Patients with cam FAIS were identified and evaluated by a board-certified orthopaedic surgeon (S.K.A, T.G.M) from the University of Utah Orthopaedic Center Clinic, as described previously [5]. Cam deformity in positive controls was identified by measuring the alpha angle from anterior-posterior (AP), 45° Dunn lateral, and modified false-profile radiographs (Table 1). Lack of cam deformity was confirmed in negative controls via alpha angle measurement from AP radiographs and oblique-axial reformatted MRI [5]. Alpha angles >55.5° from any viewing perspective indicated a cam deformity [40] (Fig. 1). Plain-film radiographs were also used to exclude individuals with acetabular dysplasia, Legg–Calvé–Perthes disease, slipped capital femoral epiphysis, and radiographic evidence of hip OA (Tönnis grade > 1).

Table 1.

Overview of screening and collection differences within study population.

Study	Cohort	Screening	MRI Purpose	CT Purpose	BVR Imaging
Newly recruited	n = 7 positive controls n = 9 patients	AP, 45° Dunn lateral, and modified false-profile radiographs	Hip/knee surface reconstruction PI/SS/femoral version measurement	---	Pulsed fluoroscopy (Imaging Systems & Services Inc.)
Atkins et al. [5,39]	n = 10 negative controls n = 6 patients	AP radiographs, oblique-axial reformatted MRI	PI/SS measurement	Hip/knee surface reconstruction Femoral version measurement	Continuous fluoroscopy (Radiological Imaging Services)

AP: anterior-posterior, MRI: magnetic resonance imaging, PI: pelvic incidence, SS: sacral slope.

Fig. 1.

Example alpha angle measurement of cam lesion in FAIS patient. Oblique axial reformatted view from axial MRI shown.

2.2. Volumetric Imaging

Participants received hip and knee volumetric imaging (Table 1). A pelvis and knee CT scan captured the full pelvis, proximal femur, and knee of previously recruited participants [5]. Previously recruited participants had also received a T1-weighted, gradient-echo MRI acquisition to visualize sacral alignment. The hip CT was not used for sacral alignment measurement since imaging was done under traction to visualize hip articular cartilage structures for a separate study. To limit radiation exposure, newly recruited participants did not receive a CT scan (Table 1). Instead, they were imaged with a 3D T1-weighted spoiled multi-echo gradient echo MR acquisition (3T Siemens Prisma, voxel size = 0.5x0.5x0.8 mm, 768x768x320 matrix; Siemens, Washington, D.C., USA) to capture the full pelvis and proximal femurs [41]. MR acquisitions on newly recruited participants were subsequently converted to a synthetic CT (sCT) (BoneMRI v1.6, MRIGuidance; Utrecht, Netherlands) [41]. Initially, a sCT was not available from BoneMRI for the distal femur. As such, all positive controls and three of the newly recruited patients were imaged with a modified Volume Interpolated Breath-hold Examination (VIBE) MR scan (voxel size = 0.9x0.9x0.9 mm, 448x224x228 matrix) to capture distal femur anatomy. However, the sCT reconstruction for the knee became available midway through data collection and was used in-lieu of the VIBE MR scan on six newly recruited patients to offer more efficient distal femur segmentation and reconstruction. Using sCT to reconstruct femur and pelvic anatomy and to measure in vivo hip kinematics from BVR data was recently shown to provide similar accuracy and reliability as a conventional CT scan [42].

2.3. Anatomy Reconstruction and Coordinate System Generation

The full pelvis and side of interest proximal and distal femur were semi-automatically segmented (Corview v1.32.2, Marrek Inc.; Sandy, UT, USA). Segmentations were manually refined, and surfaces were generated, smoothed, and decimated (Amira v6.0.1, FEI; Hillsboro, OR, USA). CT data was used for surface reconstructions for previously recruited participants, while MR-derived images (sCT or VIBE) were used for surface reconstructions of newly recruited participants (Table 1).

Landmarks of the femur and pelvis were semi-automatically identified using a previously described process [5,42]. Briefly, curvatures of the distal femur and pelvis were isolated to guide the selection of femoral epicondyle and anterior/posterior superior iliac spine landmarks (FEBio Studio 1.6.1, febio.org; Salt Lake City, UT, USA). The femoral and acetabular joint centers were identified as the center of a best-fit conchoid [43,44] of the isolated femoral head and lunate surface of the acetabulum, respectively. Anatomical coordinate systems were defined such that the x-axis pointed anatomically right, the y-axis pointed anteriorly, and the z-axis pointed superiorly.

2.4. Morphological Comparison

The side of interest proximal femur and pelvis surfaces were used to generate femur and pelvis particle-based correspondence models (ShapeWorks v6.6.0, shape-works.sci.utah.edu; Salt Lake City, UT, USA). The proximal femur correspondence model was configured based on a previously described approach [32], except the number of correspondence particles was increased from 1,024 to 2,048 to improve model accuracy. As done for the femur, pelvis surfaces were reflected if left-sided and aligned via an iterative closest point algorithm. Given its size and complex geometry, the pelvis was sampled with 4,096 correspondence particles. Prior to optimization, landmarks were placed superior to the acetabulum and near the posterior superior iliac spine to help correctly identify the anterior and posterior sides of the pelvis.

Femoral version, pelvic incidence (PI), and sacral slope (SS) were measured by a board-certified musculoskeletal radiologist (M.K.M) to gain further insight into femur morphology and sacrum alignment (Philips IntelliSpace PACS; Amsterdam, Netherlands). Femoral version was measured from volumetric data of the proximal and distal femur [45] (Table 1). PI and SS were measured from supine MRI images [46] (Table 1). While sacral alignment measurements can change from standing to supine [47], the relative relationship between supine sacral alignment measurements from MRI was shown to have strong to very strong correlations to conventional standing radiographic measurements [46].

2.5. Biplane Videoradiography Imaging and Kinematic Analysis

The hip of the side of interest was imaged with one of two validated BVR systems (Table 1). Previously recruited participants were imaged under continuous fluoroscopy, with maximum energy settings of 100 kVp, 457 mA, and 7.0 ms exposure [5,39]. Newly recruited participants were imaged with a pulsed BVR system, with maximum energy settings of 100 kVp, 125 mA, and 1.2 ms exposure (Imaging Systems & Services Inc.; Painesville, OH, USA). Energy settings less than the maximum were often used when sufficient BVR image quality could still be achieved. The neutral standing posture (i.e., ‘static’) of all participants was captured. Participants were instructed to face forward while in a comfortable stance. Up to three trials were obtained to ensure sufficient image quality and positioning; however, only a single trial was tracked to determine the orientation of the pelvis and hip joint.

A validated model-based tracking approach [35,42] was used to identify the proximal femur and pelvis poses for at least twelve frames of the static trial. Landmark locations were used to construct femur and pelvis anatomical coordinate systems for each tracked frame of the static trial. Hip joint orientation was defined as the rotation matrix of the femoral coordinate system with respect to the pelvic coordinate system. Orientation of the pelvis was generated similarly with respect to the lab coordinate system. A representative hip joint and pelvis orientation was calculated by averaging rotation matrices across frames of the static trial using quaternions [48] for each participant. Hip joint angles were computed from the average rotation matrix using a x-y-z Cardan decomposition. Flexion, abduction, and external rotation were defined as positive. For pelvic standing orientation, the sagittal and coronal plane components of the angle between the lab vertical axis and pelvic superior-inferior anatomical axis were used to compute pelvic tilt and obliquity, respectively. The transverse plane component of the angle between the lab forward direction and the pelvic anterior-posterior axis were used to measure pelvic rotation. Anterior pelvic tilt, side of interest upward obliquity, and pelvic external rotation were defined as positive.

2.6. Statistical Analysis

Anthropometric, femoral version, PI, and SS group differences were assessed between patients, positive controls, and negative controls using two-tailed student’s t-tests.

Morphological variations of the proximal femur and pelvis were evaluated across the population and between experimental groups using methods previously described elsewhere [32]. Briefly, principal component analysis (PCA) and linear discriminant analysis (LDA) were performed separately for the proximal femur and pelvis models. PCA helped identify the major morphological variation of the femur and pelvis shape within the population. A parallel analysis identified the number of PCA modes to retain for further analysis. Within each of these retained modes the principal component scores between groups were compared using a one-way ANOVA upon confirmation of normality via a Shapiro-Wilk test. A Kruskal-Wallis test assessed group differences if the PCA contribution within a group was not normally distributed. Comparing the component scores aided in detecting differences in shape variation between groups captured by each component. LDA was used to distill the high-dimensional data of the shape models to a single value or shape score (SScore). LDA is a dimensionality reduction technique that finds a linear combination of features that maximizes separation between different classes within the data. In this case, for each bone, an LDA projected the shape data on a linear discriminant normalized scale between the mean negative control shape (-1) and the mean patient group shape (+1). LDA was performed to provide femur and pelvis SScore values for the population to quantify the similarity of a participant’s bone shape compared to group mean bone shapes. Two-tailed student’s t-tests were used to compare mean SScores of the positive control group to the negative control and patient groups. Lastly, regional differences between group mean shapes were compared by visualizing colormaps of the Euclidean distance between group mean surfaces.

Hip joint angles and pelvis orientation while standing were compared between patients, positive controls, and negative controls. Normality within each group was confirmed for each variable using a Shapiro-Wilk test. Two-tailed student’s t-tests were then used to compare the group mean 3D hip and pelvis standing postures with Benjamini-Hochberg correction [49] to account for false discovery across the six tests. Multiplicity correction was applied only across comparisons for each degree of freedom, as the primary interest was in individual groupwise differences rather than overall differences among all three groups.

Lastly, correlations were assessed between femur and pelvis SSM metrics and standing posture. Specifically, correlations were assessed between each hip and pelvis rotational degree of freedom and an individual’s component score within a retained PCA mode. Correlations between standing posture variables and an individual’s femoral or pelvis SScore were also evaluated. Pearson’s correlation coefficient (r-value) was used to interpret the strength of the correlation as negligible (0.00-0.30), weak (0.31-0.50), moderate (0.51-0.70), strong (0.71-0.90), or very strong (0.91-1.00) [50]. Orientation calculations and statistical analysis were performed in MATLAB.

2. Results

The mean age and BMI of the patient group was greater than the negative control group (Table 2). Femoral version, PI, and SS yielded no significant differences between groups.

Table 2.

Population demographics and mean (±standard deviation) radiographic measures.

	Patients	Positive Controls	Negative Controls
Side (R/L)	9 / 6	6 / 1	6 / 4
Sex (M/F)	8 / 7	3 / 4	5 / 5
Age (years)	29 (6.6)*	24 (2.9)	23 (2.3)
BMI (kg/m2)	24.3 (3.4)*	22.8 (2.7)	21.1 (2.0)
Femoral Version (°)	12.0 (5.9)	11.3 (4.8)	14.7 (6.7)
Pelvic Incidence (°)	50.5 (10.2)	52.3 (15.8)	49.5 (9.7)
Sacral Slope (°)	41.8 (7.9)	41.9 (11.6)	44.3 (8.8)

^*Significantly different from negative controls

Parallel analysis identified five retained modes for the femur and six retained modes for the pelvis, accounting for 72.9% and 81.0% of the cumulative variance, respectively. Femur and pelvis group comparisons of the PCA component scores within each retained mode yielded no statistically significant differences.

In negative controls, observed regional differences between group mean femur surfaces highlighted less prominence around the anterosuperior femur head-neck junction (i.e., cam region) (Fig. 2, top). Positive controls had a more inferior lesser trochanter and more pronounced intertrochanteric region than patients and negative controls. Patient and negative control group mean pelvis shapes were similar; however, the posteroinferior and superior acetabular rim was more pronounced in patients (Fig. 2, bottom). Positive controls had a more distinct pelvis shape than the two other groups, including a more internally rotated ilium and medialized ischium. Additionally, the anterosuperior acetabular lunate surface of positive controls was more prominent compared to patients, which resulted in a more retroverted orientation of the acetabulum. The positive control group had different mean femur SScores (0.36±0.88, P<0.01) and mean pelvis SScores (0.48±0.89, P<0.01) from the negative control group (femur: −1.0±0.96; pelvis: −1.0±1.49). There were no SScore differences between positive controls and patients (femur: 1.0±1.68; pelvis: 1.0±1.43).

Fig. 2.

Regional differences between participant group mean femurs and pelvises. Negative control femurs had a less prominent cam region, and positive controls had a more inferior lesser trochanter (top). Positive control pelvises had a more internally rotated ilium and medial ischium. Patients and negative controls had less extreme pelvis shape differences. The surface shown is the first group listed in the column headers with colormaps indicating movement from the second listed surface to the first. Red and blue in colormaps indicate where the first group was more and less prominent compared to the second group, respectively.

In the evaluation of hip and pelvis standing posture, patients stood with more hip extension and less anterior pelvic tilt when compared to negative controls (Fig. 3). No other significant differences in hip and pelvis standing posture were found between groups.

Fig. 3.

Group mean standing hip and pelvic posture for the three participant groups. Patients stood with less anterior pelvic tilt and greater hip extension than negative controls on average. Brackets indicate mean ± standard deviation. Asterisks indicate significance between groups after Holm-Sidak correction for multiplicity (P<0.05).

Three significant correlations were identified between retained PCA modes in the femur and standing posture variables. PCA mode 1 displayed a weak, negative correlation with anterior pelvic tilt (Fig. 4, top left). Positive mode 1 values primarily involved a more anterosuperior lesser trochanter and described 39.0% of the variance in the population (Fig. 4, bottom left). Mode 2 was moderately, negatively correlated with pelvic upward obliquity (Fig. 4, top right). Mode 2 also displayed a weak, positive correlation with hip abduction. Positive mode 2 values corresponded with a more pronounced cam region and elevated greater trochanter and explained 12.4% of the variance (Fig. 4, bottom right). Femur SScore also displayed weak, negative correlation with hip flexion and anterior pelvic tilt (Fig. 5).

Fig. 4.

Significant correlations between variations in proximal femur morphology and standing posture across the study population. Decreased anterior pelvic tilt was significantly correlated with femur PCA Mode 1 (left). Positive Mode 1 scores corresponded to a more anterosuperiorly positioned lesser trochanter. Greater pelvic down obliquity and hip abduction (not shown) were significantly correlated with femur PCA Mode 2 (right). Positive Mode 2 scores corresponded to a more pronounced cam region and elevated greater trochanter. Femur shape reconstructions at ±2 SD along modes are shown from a posteromedial (top) and anterolateral (bottom) view, with red and blue indicating displacement toward and away from the population mean shape, respectively.

Fig. 5.

Association between a single value metric describing proximal femur shape and standing pelvic tilt. Decreased anterior pelvic tilt and greater hip extension (not shown) were significantly correlated with more patient-like femur shape score. Group distributions of shape scores are shown on the x-axis. Mean positive control shape score was different from negative controls but not patients. Significant differences are indicated with asterisks (*P<0.05).

Two retained PCA modes for the pelvis yielded three significant correlations to standing posture variables. PCA mode 1 scores had a weak, negative correlation with anterior pelvic tilt (Fig. 6, top left). Mode 1 also had a weak, negative correlation with hip flexion. Positive mode 1 scores mainly corresponded to more anterosuperior acetabular coverage, a medialized ischium, and an internally rotated ilium while explaining 32.7% of the pelvis model variance (Fig. 6, bottom left). Pelvis PCA mode 3 displayed a weak, negative correlation with hip abduction (Fig. 6, top right). Positive mode 3 values consisted of a more laterally convex pelvis shape and explained 12.4% of the variance (Fig. 6, bottom right). Pelvis SScore was not significantly correlated to any standing posture variable.

Fig. 6.

Significant correlations between variations in pelvis morphology and standing posture across the study population. Decreased anterior pelvic tilt and greater hip extension (not shown) were significantly correlated with pelvis PCA Mode 1 (left). Positive Mode 1 scores corresponded to more anterosuperior acetabular coverage, a medialized ischium, and an internally rotated ilium. Decreased hip abduction was significantly correlated with pelvis PCA Mode 3 (right). Positive Mode 3 scores corresponded to a more laterally convex pelvis shape. Pelvis shape reconstructions at ±2 SD along modes are shown from a lateral (top) and medial (bottom) view, with red and blue indicating displacement toward and away from the population mean shape, respectively.

3. Discussion

This study investigated the associations between hip morphology and standing posture using SSM and BVR in individuals with and without cam morphology and symptoms. We identified that patients with cam FAIS had similar hip and pelvis standing posture compared to positive controls but stood with less anterior pelvic tilt and greater hip extension than negative controls. We also identified significant correlations between femur and pelvis PCA modes of variation and kinematic variables, including pelvic tilt, revealing features to target for multivariate analysis techniques. Further, patients and positive controls displayed shape differences around attachment sites of common hip crossing muscles on the femur and pelvis. Positive controls displayed a unique pelvis shape with an internally rotated ilium, potentially broadening the sacroiliac space and highlighting a potential mechanism of symptom-avoiding compensation through greater spinopelvic mobility. The results herein support prior research emphasizing the potential role of pelvic tilt in the symptomatology of FAIS and describe novel associations between hip morphology and standing posture using independent measurement techniques. Future work should increase samples sizes and incorporate analysis involving dynamic activities to support multivariate approaches for phenotyping FAIS symptomatology.

This study demonstrated that patients with FAIS stood with significantly less anterior pelvic tilt than negative controls (Fig. 3). Pelvic tilt magnitude and range of motion have previously been shown to differ between these populations during dynamic activities [5,7,9,10,20]. The use of measurement techniques prone to error [7,9,10,20] and small sample sizes [5] may have prevented prior studies from observing similar differences in postures that may avoid cam impingement (i.e., standing). Interestingly, excessive anterior pelvic tilt while standing is commonly deemed an impairment when performing a visual assessment of hip function [48,49], despite patients with FAIS in this study displaying more posterior pelvic tilt. These results highlight that patients may preemptively decrease anterior pelvic tilt to facilitate a greater hip range of motion dynamically [11]. Our findings justify further investigation into defining appropriate functional assessments for this population.

Positive controls did not display differences to either group in standing posture or sacral alignment, limiting insights into the spinopelvic complex that may be insightful to understanding symptom-avoiding mechanisms. Femur and pelvis SScore results also showed that positive control femur and pelvis shape was not different from patients. Assessing SScore differences was an effective approach for assessing the similarity of the patient and positive control femur morphology, but it may be misleading for assessing the similarity of pelvic morphology. The pelvis LDA was generated without known morphological differences defined between endpoint shapes, unlike the femur LDA. Regional differences between group mean pelvis shapes (Fig. 2) highlight that patients and negative control mean pelvis shapes were quite similar. Ultimately, the pelvic LDA indicates that positive controls have a pelvic shape more like the patients than the negative controls, but further objective analysis of pelvic shape differences between these populations is warranted.

Regional variations in positive control morphology could be used to infer morphology that promotes FAIS symptom avoidance in individuals with cam morphology of the proximal femur. Positive controls, on average, had an internally rotated ilium and more retroverted acetabular orientation, a pelvis shape that contradicts associations previously seen in patients with FAIS or dysplasia [9]. Also, pelvic PCA mode 1 scores, partially describing internal ilium rotation, were correlated with reduced anterior pelvic tilt. An internally rotated ilium could allow greater space for the sacrum to move, potentially increasing the ability to generate a posteriorly directed pelvis tilt. This theory aligns with prior work observing how increased spinopelvic mobility is associated with less severe [6] or non-existent [7,10] FAIS symptoms despite cam deformity. Moreover, morphological differences at muscle attachment sites in positive controls suggest adaptations in muscles that facilitate impingement-avoiding movements [51], such as posterior pelvic tilt, hip abduction, and external rotation. In comparison to patients, positive controls have been previously shown to have greater isometric [20] and functional [52] hip extension and flexion strength, which would align with the muscle groups involved in controlling pelvic tilt. However, the isometric activities observed [20], and generic bone models used [52] in these prior studies may have limited greater insights into muscle strength differences. In the future, it will be important to assess associations between SSM and BVR during dynamic activities to establish the role of the sacrum in symptom avoidance. Further, musculoskeletal modeling studies should consider participant or group-specific bone geometries to account for the distinct morphology of the positive controls.

The present study had numerous associations between femur and pelvis shape metrics and standing posture (Fig. 4-6). Such associations help identify anatomical features that facilitate or necessitate specific pelvic orientation. Moreover, these associations highlight variables of interest for multivariate clustering methods that could phenotype whether aspects of morphology or orientation are driving symptoms in a specific patient. Clinicians could leverage these relationships to determine the best treatment approach for a patient, whether that is established arthroscopic removal of the cam lesion [53], conservative treatment increasing muscle strength and hip mobility [54], or new approaches aimed at altering spinopelvic alignment. However, while these associations can be informative, none of the significant correlations involved PCA modes that were different between experimental groups. Therefore, a clear symptom-inducing association between specific anatomical features and hip/pelvis orientation remains unknown. Further, most of the significant correlations were weak. Additional differences in group mean PCA mode component scores and form-function relationships may be found in future studies with larger sample sizes. Still, the correlations identified could be leveraged in the future for more robust multivariate analyses that are designed to improve clinical decision-making.

This study had limitations that warrant discussion. Anatomical landmarks were selected to define patient-specific anatomical coordinate systems for hip orientation measurements from BVR. Landmarks used for coordinate system creation follow international standards [55] but methods to select landmarks could influence measured standing orientation results. The consistent and repeatable [35] landmark selection used herein bolsters the group-wise differences found but values of pelvic tilt need to be cautiously compared between studies with different methods. Additionally, approximately half of the study population was recruited as part of a prior study, where cam morphology in the asymptomatic participants was identified using slightly different radiographic views of the femur. Still, this limitation is offset by the fact that we used SSM to objectively characterize and compare hip anatomy. Importantly, we demonstrated that the mean femur SScore of the positive control group was statistically different from the negative controls. There were positive control participants and patients with negative femur SScores since the LDA to generate SScores incorporates the entire femur shape, not just the cam region. Regional differences between mean femur shapes also highlight that negative controls had a much less prominent cam region than the other groups. While all groups had overlapping distributions of proximal femur shape, observable differences in the mean shapes existed amongst the groups. We also acknowledge that positive controls could develop FAIS symptoms later in life, but all were asymptomatic at the time of data collection. Longitudinal studies will be required to establish the specific risk factors that distinguish individuals with asymptomatic cam morphology from those who will progress to symptomatic FAIS.

4. Conclusion

This study describes a novel approach incorporating accurate, reliable, 3D in vivo hip and pelvis orientation measurements from BVR with objective 3D evaluation of hip morphology using SSM to improve cam morphology form-function relationships. Our results suggest that patients with FAIS adopt posterior pelvic tilt while standing but may have pelvic morphology that limits further posterior pelvic tilt dynamically, potentially leading to FAIS symptoms. Correlations between hip morphology and standing posture support further investigation into identifying what links spinopelvic mobility to FAIS symptoms. Also, regional pelvic shape differences highlight the importance of considering patient-specific morphology for future analysis of muscle strength and function in this population. Findings from this study establish the framework to phenotype FAIS symptomatology to enable more informed patient-specific treatment strategies of FAIS, potentially improving clinical outcomes.