Pathomechanics of

FAI and Hip Dysplasia: Current Clinical and Translational Science Perspectives
Michael D. Harris, BS Christopher L. Peters, MD Jill Erickson, PAC Corinne R. Henak, BS Ashley L. Kapron, BS Christine L. Abraham, BA Jeffrey A. Weiss, PhD Andrew E. Anderson, PhD

University of Utah, Salt Lake City, Utah

Please see Conflict of Interest Disclosures as listed in the AAOS program and AAOS website. All studies approved by U of U IRB.

Introduction

Structural hip disorders such as femoroacetabular impingement (FAI) and dysplasia are risk factors for the development of hip osteoarthritis (OA). The clinical presentation of FAI has been qualitatively described and correlated with 2D radiographic measurements. However, the tremendous heterogeneity present in FAI patients cannot be captured with 2D metrics; 3D methods are clearly desired. Hip pathomechanics (kinematics, muscle forces, cartilage contact mechanics) are ultimately responsible for tissue damage. Oversimplified 2D measurements of anatomy and clinical/surgical observations alone cannot explain the mechanical pathogenesis of chondrolabral injury and OA in hips with structural disorders. An improved understanding of structural hip disorders requires one to relate 3D hip pathoanatomy to the underlying pathomechanics; few studies exist in this regard.

Specific and thorough characterization of subject specific 3D hip anatomy and pathomechanics can explain characteristic damage patterns. Relegating these higher-ordered data (subject-specific 3D anatomical and mechanical data) to standard clinical metrics could identify those current measurements with the best likelihood of describing underlying damage. Higher-ordered data could also yield novel measurements that are objective, discriminatory, and/or provide descriptions of what should be considered normal variation in hip anatomy. This exhibit presents methods and experiments we have used within the past decade to quantify subject-specific and group-wise 3D hip anatomy, kinematics, kinetics, muscle forces and cartilage/labrum mechanics.

Clinical Motivation
Clinical and surgical observations of 420 hips treated with open surgical procedures for FAI (213) and dysplasia (207) were prospectively collected from 1996-2011. Chondrolabral injury patterns were recorded for each. Preoperative and postoperative pain and function were assessed prospectively with the modified Harris Hip Score (HHS). Our practices continue to see an increase in the number of patients presenting with structural disorders of the hip.

Structural Disorder
Dysplasia 80% classic 20% retroversion FAI 55% combined 40% cam

Number
207

Mean HHS improvement

25 points

Intraoperative Findings
85% decreased femoral head neck offset 21% abnormal labrum 5% other 98% decreased femoral head neck offset 85% abnormal labrum 75% acetabular hyaline cartilage lesions

Failures
7.8%

213

24 points

6%

Basic Science Overview

Mission Statement
Combine imaging/3D reconstruction, motion analysis, and computational biomechanics to address important problems related to femoroacetabular impingement and acetabular dysplasia (see Venn diagram).

Motion Analysis

Motion analysis equipment includes 10-camera Vicon system, custom High-speed Dual-Fluoroscopy System and Instrumented Treadmill. Vicon System tracks markers attached to skin.
- Joint angles, stride lengths, etc

Projects may involve single approach - Femoral head asphericity (Imaging/3D Recon.) Most studies combine two approaches - FEA (3D Recon /Comp. Biomech.) - Dual-fluoroscopy (Imaging/Motion Analysis) Ongoing work seeks to combine all to enable patient-specific analyses - FEA models driven with patient kinematics and muscle forces
Imaging & 3D Recon Motion Analysis

Dual Fluoroscopy system tracks in-vivo joint kinematics to sub-millimeter and sub-angle accuracy4 (Fig. 2).
- Image intensifier and emitter mounted on separate carts to facilitate imaging supine, prone, standing - High-speed digital cameras mounted at output phosphor window - Model-based tracking used to quantify in-vivo bone motions (without implanting beads into subjects)

Computational Biomechanics

Bertec Instrumented reaction forces (Fig. 3).

Treadmill

measures

ground

Imaging and 3D Reconstruction

- Belts 20 wide (facilitate running on a single belt) - Up to 6 m/s (sprinting speed) - Inclines to 15

CT arthrography provides base imaging modality for most studies at Utah (>70 FAI/dysplasia patients and 25 volunteers to date).
- Fast acquisition times, high spatial resolution, and excellent delineation of bone and cartilage (Fig. 1) - Intra-articular contrast injected under fluoroscopic guidance (15-25 cc of Omnipaque to Lidocaine [1:2]) (IRB in place for >8 yrs) - Hip joint traction applied with a hare-traction splint during CT scan (poor quality without traction)
Figure 2. Utah highspeed dual fluoroscopy system. Figure 3. Bertec dualbelt instrumented treadmill at Utah Motion Lab.

Computational Biomechanics

Focus of Utah group has been finite element (FE) analysis, with ongoing work in statistical shape and musculoskeletal modeling. FE protocol requires several steps (Fig. 4).
- Numerous validation studies1,5

CTA

Volumetric Image Data

MRA

Segmentation

Figure 1. Radiographs, axial and coronal CT arthrography slices, and 3D recons of labrum and cartilage. Traction ensures adequate spacing of cartilage. Top row moderately dysplastic hip. Bottom row severely dysplastic hip with large labrum.

Discretization Hex Tet

FE models include patientspecific cartilage and bone contact geometry.

- Idealized contact geometry does not provide accurate predictions

Model Geometry

Utah hip CT arthrography protocol has been extensively validated.1-3

- Accuracy of cortical bone thickness1 - Influence of joint spacing and dilution of contrast agent2 - Accuracy of cartilage thickness3

SUBJECT-SPECIFIC
COMPUTATIONAL MODEL

Validation of Model or Protocol

Bone and cartilage semi-automatically segmented into 3D reconstructions using Amira

- Used to quantify femoral head asphericity, perform statistical shape modeling, create 3D finite element models

Cartilage contact stresses in normal hips6, traditionally dysplastic hips, and hips with retroversion. Future FE models will include patient-specific kinematics and kinetics.
- Dual-fluoroscopy with treadmill - Musculoskeletal models predict muscle forces for FE models

Literature Data Boundary and Loading Conditions Constitutive Assumptions Experimental Data Image Data

Literature Data

Motion Analysis

Digitally reconstructed radiographs generated using Amira

- Enable creation of views from any perspective of interest under highly controlled parameters

Figure 4. Flowchart of steps involved with developing an FE model. Adapted from 7.

References [1] Anderson AE, et al. J Biomech Eng, 127(3) 2005. [2] Anderson AE, et al. Radiology 246(1) 2008. [3] Allen BC, et al. Radiology 255(2), 2010. [4] Bey MJ, et al. J Biomech Eng, 128:604-609, 2006. [5] Anderson AE, et al. J Biomech Eng 130(5) 2008. [6] Harris MD, et al. J Orthop Res, 30(7) 2012. [7] Henak CR, et al. J Biomech Eng, 2013.

Anatomy
Motivation to Study FAI Anatomy

3D measures of femoral asphericity and statistical shape modeling of femur morphology can (1) objectively assess the spectrum of cam FAI deformities, (2) improve presurgical planning, and (3) refine 2D clinical screening protocols.

Digitally Reconstructed Radiographs (DRR) used to simulate x-rays from 5 common views of femurs (Fig. 6) Alpha angles measured on DRRs and radial CT correlated with 3D maximum deviation from spheres

Figure 6. Top From CT, DRRs can be created and rotated to simulate any radiographic projection view; Coupling with 3D reconstructions allows isolation of region of interest (i.e. femur). Bottom DRR and radial CT examples.

3D Measures of Femoral Asphericity

Objective isolation of femoral head (Fig. 5)

3D reconstruction from volumetric CT images Head-neck junction mathematically determined using 1st principal curvature

30 60

Sphere-fitting and 3D asphericity

Isolated head fit to sphere and 3D deviations between sphere and native femoral head calculated Maximum deviation and protrusion area (i.e. sections of femoral head protruding above sphere) calculated Location and magnitude of maximum deviation and protrusion areas compared between control and cam FAI subjects
3.0 mm
AP Frog -leg

45 Dunn (neutral rotation) 90 Radial / Oblique Axial

90

30

60

90

45 Dunn (external rotation)

Significance

Crosstable lateral

best-fit sphere native femur 0.01 mm

Figure 5. Femoral head isolation and sphere fitting. Left Femoral head delineated from neck using inflection points around circumference of head neck junction (black line). Middle Isolated head (offwhite) projected onto best fitting sphere surface (green). Right Deviations (mm) between femur and bestfit sphere calculated across isolated surface of head.

45 Dunn had strongest correlation to 3D asphericity when femur was externally rotated; 45 Dunn with neutral rotation was weakly correlated. Thus patient positioning during screening is important for detecting cam FAI AP had weakest correlation despite commonly being first clinical xray 60 radial CT strongly correlated to 3D asphericity, but 30 was weak; need to measure alpha angle from multiple radial views even in anterosuperior region of head

Statistical Shape Modeling (SSM)

SSM can objectively identify and quantify shape variations without any assumption about geometry6. Correspondence particles placed on 41 control, 30 cam FAI femurs and used to generate average cam and control femur shapes and define areas of inter-/intra-population variation.

Application to cam FAI1

15 cam FAI patients, 15 matched controls Maximum deviations from best-fit sphere: Controls = 2.41 0.31 mm, Patients = 4.99 0.39 mm Surface area of protrusions: Controls = 827.1 42.2 mm2, Patients = 675.8 39.3 mm2

Significance
Femur shape variation greatest in femoral offset, greater trochanter height, and head-neck junction Average cam femur protruded above average control by max of 3.3 mm with sustained protrusions of 2.5-3.0 mm along anterolateral head-neck junction SSM may be useful for developing new measurements of pathological anatomy and can provide templates to guide patientspecific surgical resection of bone.
6 mm

Significance
First objective, quantitative 3D description of where and how much control and FAI femurs deviate from spheres Control femurs have low-lying, broad protrusion; patient protrusions are more abrupt, more concentrated Amount of normal asphericity should be considered when treating cam FAI

Alpha Angle and 3D Asphericity

Strength of alpha angle to detect 3D asphericity needs to be objectively assessed.

Alpha angle measurements can vary widely from different views of same femur2,3 Previous alpha angle assessments revert to 2D reference standards or quantify only repeatability of measurements3,4,5
0 mm

Figure 7. Color plot description of where and how much a single cam femur deviates from the average control femur. Such plots may be used for surgical planning to guide resection of cam lesions back to a statistically determined normal shape.

References [1] Harris et al., (2013) Ann Biomed Eng. (in press). [2] Meyer et al., (2006) Clin Orthop Rel Res. 445:181-5. [3] Domayer et al., (2011) Eur J Radiol, 80(3):805-10. [4] Carlisle et al., (2011) Iowa Orthop J, 31:52-8. [5] Rakhra et al (2009) Clin Orthop Rel Res. 467(3)660-5. [6] Cates et al., (2007) Inf Proc Med Imaging. 20:333-345.

Kinematics & Kinetics

Motivation
Chondrolabral damage in hips with femoroacetabular impingement (FAI) may result from motion conflict due to abnormal bony morphology. However, hip impingement has not been quantified in vivo. Abnormal morphology can reduce hip range of motion1. Muscle mechanics may also be affected but muscle forces and joint reaction forces in pathologic hips are unknown and cannot be measured in vivo.

Musculoskeletal Modeling

In Vivo Kinematics

Musculoskeletal modeling couples kinematics from fluoroscopy and/or skin markers and force data with computational representations of muscles (Fig. 10). Model uses experimental input data, known muscle behavior, and optimization theory to determine forces generated by individual muscles6. Feasibility study complete to validate model estimations of muscle activation (Fig. 10). Model applied to 7 healthy subjects and 7 dysplasia patients during normal gait. Angles and moments at pelvis, hip, and knee calculated with inverse kinematics and inverse dynamics. Muscle and joint reaction forces determined using static optimization7.

Figure 8 left. Subjects left hip flexed for impingement exam in DFS. Middle, right. Fluoroscopy images of impingement exam with aligned DRRs.

Normal male volunteer was imaged in the DFS (Fig. 8, left) and CT scanner. An orthopaedic surgeon moved subjects hip during 3 clinical exams. Model-based tracking software determined bone position and orientation2 (Fig. 8, middle, right). Raw kinematics converted to joint angles and translations3,4. Custom PostView5 tool visualizes bone motion and calculates the distance between bones.

Figure 10. Left OpenSim model with 96 muscle actuators (representing 80 muscles). Right upper Muscle activations from EMG (in vivo) compared to OpenSim (model prediction) for hamstrings. Right lower Joint reaction forces for control and dysplastic subjects during gait. Vertical bars indicate standard deviations.

Joint angles, moments, muscle forces in agreement with literature values8-10. Dysplastic patients had significantly lower joint reaction force peak at post-heel-strike loading response. Majority of muscle forces not significantly different between groups. More extreme ranges of motion likely required to elicit more substantial kinematic differences, and subsequent muscular compensations.

Impact and Future Direction

Dual fluoroscopy and model-based tracking could provide substantial insights into how FAI alters hip kinematics and causes chondrolabral damage. - This protocol is actively being applied to a cohort of normal volunteers and FAI patients. Musculoskeletal modeling can detect/estimate muscle weaknesses evidenced as altered kinematics. Joint reaction forces and joint angles can be used to improve specificity of chondrolabral contact models for improved surgical planning and evaluation.

Figure 9. Joint angles during clinical exams on live subject with normal hip anatomy (left column). Bonebone distance is displayed on pelvis (middle column) and femur (right column) at time points indicated in joint angle plots by vertical grey line.

References [1] Lamontagne et al., JBJS 2011. [2] Bey et al., (2006) J Biomech Eng, 128:604-609. [3] Wu et al., (2002) J Biomech, 35:543-548. [4] Grood and Suntay (1983) J Biomech Eng, 105:136-144. [5] PostView. http://mrl.sci.utah.edu/software.php [6] Delp et al, IEEE Trans Biomed Eng 54(11) 2007. [7] Anderson and Pandy, J Biomech 34(2) 2001. [8] Lui et al, J Biomech 41(15), 2008. [9] Lewis et al., Gait & Post, 32, 2010. [10] Oberg et al., J Rehab Res & Dev, 31(3), 1994.

Mechanics in Dysplastic Hips

Labrum Load Support (%)

Methods

Subjects were recruited and imaged with IRB approval: - Ten subjects with normal hip anatomy and no history of hip pain (5 male, LCEA 33.5 5.4)5. - Ten subjects with hip pain secondary to acetabular dysplasia (3 male, LCEA 14.8 4.6). Subject-specific finite element models were generated (Fig. 11). The percent of the total load supported by the labrum was evaluated. Contact stress and contact area were evaluated in the anterior, superior and posterior acetabular labrum. Contact stress and contact area Figure 11. Subjectspecific finite were evaluated on the element model of the bones acetabular cartilage. (white), femoral cartilage (blue), cartilage (yellow) and Results compared between acetabular acetabular labrum (red). Lines groups using t-tests indicate element boundaries. (significant at p 0.05).

Impact

0 MPa Figure 13. Contact stress during WH in representative hips, displayed on cartilage and labrum (top) and on cartilage only (bottom). The dashed line indicates the boundary between cartilage and labrum. Contact patterns were subjectspecific with lateral contact in dysplastic hips and distributed contact in normal hips.

Lateral contact in dysplastic hips is supported mechanically by the acetabular labrum (Fig. 14). - The larger labrum load support and labrum contact area in dysplastic hips indicates that the labrum compensates for the shallow acetabuli. This compensation caused minimal differences in cartilage contact mechanics between the two groups. Overall, this study highlights the importance of the acetabular labrum for mechanical load-bearing in the dysplastic hip. Normal
pelvis labrum lateral

Without Labrum

With Labrum

Abnormal hip contact mechanics in dysplastic hips are thought to be the cause of early onset OA1. May be due to elevated cartilage stresses. Clinically, labra in dysplastic hips are often hypertrophied and sometimes calcified2. This may suggest a mechanical role of the labrum. Differences between normal and dysplastic hip contact mechanics are unclear. Previous studies have omitted the acetabular labrum or used idealized joint geometry3,4. The aim of this study was to evaluate both chondral and labral mechanics in dysplastic hips in comparison to normal hips.

Superior Labrum 2 Contact Area (mm )

Motivation for Dysplasia Mechanics

18 16 14 12 10 8 6 4 2 0

* *

* *

100 80 60 40 20 0

* *

WH

WM

DH

AH

WH

WM

DH

AH

Figure 12. Labrum results. Percent of the total load supported by the acetabular labrum was larger in dysplastic subjects than in normal subjects during all activities (left). Labrum contact area in the superior region was larger in dysplastic subjects during all activities (right). Normal Dysplastic 8 MPa

Dysplastic

Results
The labrum supported significantly more load in dysplastic hips than in normal hips (Fig. 12). Labrum contact area in the superior region was significantly larger in dysplastic hips than in normal hips. There were few differences in cartilage mechanics. However, where there were differences, contact stress and contact area were significantly larger in normal hips than in dysplastic hips.

cartilage

femur
0 MPa 2 MPa Figure 14. Coronal crosssectional images of a representative normal hip (left) and a representative dysplastic hip (right). Contact stress in the acetabular labrum shows labral loading in the dysplastic hip but no labral loading in the normal hip.

References [1] McWilliams, D et al., Ann Rheum Dis, 2010. [2] Klaue, K et al., JBJS-B, 1991. [3] Chegini , S et al., J Orthop Res, 2009. [4] Russell, M et al., J Orthop Surg Res, 2006. [5] Harris, M et al., J Orthop Res, 2012.

Mechanics in Retroverted Hips

Motivation for Retroverted Mechanics
Acetabular retroversion may cause early OA. However, the mechanical link between acetabular retroversion and OA is not well understood1-3. Possible mechanisms include posterior overloading due to posterior deficiency and anterior femoroacetabular impingement4. The aim of this study was to compare cartilage contact mechanics between hips with retroverted acetabula and normal hips.

Methods
Subjects were recruited and imaged (CT Arthrography) with IRB approval: - Ten subjects with normal hip anatomy and no history of hip pain (5 male, LCEA 33.5 5.4)5. - Ten subjects with a radiographic cross over sign on standardized radiographs, pain and clinical exams consistent with acetabular retroversion, and who subsequently received treatment for symptomatic acetabular retroversion (9 male, LCEA 27.8 5.5). Contact stress and contact area were evaluated in the acetabulum. Results were compared between the groups using t-tests (significant at p 0.05).

Figure 15. Average contact stress during midstance of walking (WM), heel strike during stair descent (DH) and maximum flexion during chair rise (CR). Contact patterns were influenced by subject group and kinematics.

Results

Contact in retroverted hips tended to be superior and medial, while contact in normal hips tended to be more evenly distributed (Fig. 15). Contact area in normal hips was larger than in retroverted hips in the lateral region during all activities and in the anterior region during most activities. Average cartilage contact stress was larger in normal hips than in retroverted hips in the lateral, anterior, posterior and posterolateral regions during walking. Peak cartilage contact stress was larger in retroverted hips than in normal hips in the superomedial region during walking. Cartilage contact stress also varied as a result of loading scenario (Fig. 16).

Figure 16. Crosssectional images of cartilage pressure during midstance of walking. Retroverted hips had medial and superior contact while normal hips had distributed contact.

Insights and Future Direction

The results of this study are inconsistent with the proposed mechanism of elevated posterior stresses causing damage in retroverted acetabula. Therefore, future work evaluating anterior impingement in these hips is warranted. Overall, this study demonstrates superomedial cartilage contact in hips with acetabular retroversion compared to distributed cartilage contact in hips with normal bony anatomy during activities of daily living.

References [1] Ezoe, M et al., JBJS,2006. [2] Kim, W et al., JBJS-B, 2006. [3] Giori, N et al., CORR, 2003. [4] Ganz, R et al., CORR, 2008. [5] Harris, M et al., J Orthop Res, 2012.

PAO Mechanics in Retroversion

Motivation
Treatment of acetabular retroversion is controversial: patients receive debridement of the anterior acetabular rim or posteriorly directed PAO1-3. Despite improved coverage, PAO may increase cartilage stresses due to incongruencies. This study used patient-specific and idealized/spherical FE models to quantify pre- and post-op cartilage stress in a patient treated for retroversion with a PAO.

Post-op models indicated a medial shift in contact. Patient-specific: contact area remained the same for chairrise (CR) and was reduced by ~15% in walking (WH), and descending stairs (DH). Spherical: slight increase in contact area in DH, WH and reduced contact area in CR. Average contact stress increased post-op in all models except the spherical WH model.

Figure 17. Preop hip with insufficient coverage, corrected postPAO.

Methods
Recruited one male treated for retroversion by PAO. Contact stress and contact area were evaluated in the acetabulum pre- and post-op using FE models.

Results
Figure 19. Contact stress in WH. Midsagittal crosssection (right) shows the disruption of cartilage congruency postop (gaps) and increased stresses. Red arrows indicate loading direction

Insights and Future Direction

PAO alters mechanics but may not restore to normative levels. Acetabulum does not support load as effectively medially due to horseshoe shape. Congruency between cartilage layers is disrupted by PAO. Additional models are needed to elucidate the effects of congruency, cartilage thickness, and acetabular morphology.

Figure 18. Contact stress pre and postop in models with the articular surface fit to a sphere during walking at heel strike (WH), heel strike during stair descent (DH) and maximum flexion during chair rise (CR). Contact shifted medial and contact stress was more evenly distributed postop. Red arrows indicate loading direction.
References [1] Anderson A et al., J Biomech, 2010. [2] Harris M et al., JOR, 2012. [3] Armiger A et al., Acta Orthop, 2009.

Take Home Message

Body of scientific work has laid the foundation to investigate relationships between abnormal anatomy and altered mechanics. 3D methods to study anatomy further our understanding of the disease spectrum and help to isolate unique cases. Quantification of cartilage stresses may explain damage patterns and could predict surgical outcome in the absence of long-term follow-up data.

Translating scientific data to the clinic is challenging. Larger sample sizes and streamlined methods will improve translation to the clinic. Model validation also improves clinical acceptance. We envision utilizing these methods for pre-operative planning and to quantify the efficacy of surgery.