Title: Hip Structure Analysis: Using DXA Scans to Measure Bone Strength
1Hip Structure Analysis Using DXA Scans to
Measure Bone Strength
- Thomas J. Beck, Sc.D.
- Division of Medical Imaging Physics
- Department of Radiology
- The Johns Hopkins University School of Medicine
2Presentation Goals
- What strength properties can be measured?
- What is meant by structural geometry?
- What geometric parameters are important?
- How can geometry be measured from DXA scans?
- How do age trends influence geometry (and BMD)?
- How should an effective treatment improve
geometry?
3Osteoporosis
- Reduces bone mechanical strength so that
fractures occur at certain skeletal sites with
minimal trauma - Under a specific load, bone strength is a
function of - Material Strength
- Geometry (both micro and macro scales)
4Material strength
- Determined by composition of the tissue
- Not accessible by any current non-invasive
methods - Measuring from biopsies is technically difficult
- CT methods claiming to measure material strength
actually measure bone volume fractiona geometric
effect. - Very useful way of treating CT data for FEA
models but is measures of APPARENT, not actual
material strength.
5Structural Geometry
- Age and osteoporosis mainly alter the amount and
distribution of bone tissue within bones. - These are geometric effects
- Geometry CAN be measured but accuracy and
precision can be problematic. - Most imaging methods lack resolution to discern
microarchitecture
6Whole Bone Geometry
- Linear dimensions (bending moments)
7Stress Distributions in Cross-Sections
Compressive Stress
- Force is uniformly distributed over bone surface
(marrow and other soft tissue spaces do not
contribute)
Stress is inversely proportional to bone surface
area (CSA)
8Stress Distributions in Cross-Sections
- Stress not uniformly distributed over bone
surface - Greatest at outer margins
Stress is proportional to bone surface weighted
by distance from center of mass (cross-sectional
moment of inertia - CSMI) Maximum stress is
determined by the section modulus (calculated
from CSMI)
9How can we measure geometry ofcross-sections?
- Method 1 Measure from a picture of the
cross-section acquired by CT methods - Requires specialized geometry software
10Projection of bone in cross-section excludes all
soft-tissue spaces (which do not contribute to
strength)
Method 2 Measuring from a projection of the bone
tissue in the cross-section
11Thickness profile and image of cross-section have
equal
Cross-Sectional Area (CSA)
Thickness (cm)
Distance (cm)
12DXA only projects the mineral content and
eliminates the non contributing soft tissues
Applying The Principle to DXA
- Get thickness profile by dividing pixel values by
average mineral density of fully mineralized bone
Mass thickness (g/cm2)
Distance (cm)
13DXA only projects the mineral content and
eliminates the non contributing soft tissues
Applying The Principle to DXA
- Get thickness profile by dividing pixel values by
average mineral density of fully mineralized bone
Thickness (cm)
Distance (cm)
14Measuring Geometry with DXA
- Mineral profiles are extracted from DXA images by
HSA software. - Three femur cross-sections are analyzed
15Age trends in Geometry and BMD
- The HSA method permits the same bone mass data
to be expressed in terms of BMD and geometry - HSA has been applied to gt150,000 hip scans from
DXA studies around the world - Some flavor of the kinds of results obtained
particularly what HSA tells us about bone
homeostasis (self-preserving behavior).
16The Expansion Paradox
- Long bones generally expand with age by
periosteal apposition. - Appears to be a homeostatic adaptation to bone
loss that preserves section modulus. - Reduces BMD even without bone loss
- Permits section modulus to be maintained with
less material - May ultimately be responsible for fragility
17As Diameter Is Increased, Same Section Modulus
Requires Less Material
Relative value
18Simplistic Geometry of the Aging Bone
Cross-Section
20
40
60
80
100
Age
19Implications of Homeostatic Expansion
20
40
60
80
100
Age
20Characterizing cortical instability
Strength predicted by section modulus
Section modulus overestimates strength
21Local Buckling
t
ro
ro
Buckling Ratio
t
In hollow tubes, strength loss due to local
buckling begins to occur with BR gt 10 BR can be
estimated from DXA data by the HSA method
22Essential Geometric Parameters
- CSA for resistance to axially directed loads
- Section modulus for resistance to bending loads
- Buckling ratio for stability of cortex under
compressive loads (including bending).
23The DXA Bone Mineral Mass Image
- Each pixel measures mass of hydroxyapatite (HA)
in g/cm2 summed along a straight-line path
through patient from x-ray source to detector
detector
X-ray source
24The DXA Bone Mineral Mass Image
- Pixel values actually reflect thickness of
hydroxyapatite (HA) after removal of any organic
material along the path - For example 0.6 g/cm2 of mineral is equivalent
to a 0.6 g layer of HA distributed over a 1 cm2
area (there is no area in g/cm2)
25What do we really measure with BMD?
Topographic display of pixel values in neck
region
Region length Lr fixed by software
26What do we really measure with BMD?
Region Width depends on location of bone margins
27What do we really measure with BMD?
BMD average mass thickness in g/cm2 BMC BMD
(g/cm2) x BA (cm2)
28The Meaning of a Decline in BMD
- If (and only if) BA doesnt change then loss of
BMD means - Loss of bone mineral mass
- Loss of structural strength
- Does BA remain fixed in aging bones?
29DXA Age Trends RevisitedAverage values per
decade from NHANES III
Relative Value
All values normalized to means of 20-29 y/o
30Is Expansion Homeostatic? Age trends in femur
stresses in NHANES III
- Homeostasis should keep stresses (strains) at a
constant value independent of age, size, load, or
other factors - Simulated a fall on the greater trochanter using
measured geometry from HSA. - Femoral neck stresses were computed using
engineering model with impact load a function of
body weight
31Age Trends in Femoral Neck Stress
White Women
White Men
Medial
Medial
Lateral
Lateral
Lateral
Medial
32Interpretation
- Declining BMD in aging bones doesnt necessarily
cause stresses to increase - Adaptive geometric changes counter aging bone
loss. - Adaptation is achieved in an expanding bone
diameter with progressively less bone mass - Cortices may thin to the point of local
instability under unaccustomed loads (falls).
33How should effective treatments alter femur
geometry?
- CSA should increase to reduce axial stresses
- Section modulus should increase to reduce bending
stresses - Buckling ratios should DECREASE to reduce
susceptibility to buckling
34Examples of HSA in Clinical Trials
- Raloxifene vs. Placebo (Eli Lilly Co.)
- 1365 osteoporotic women on placebo vs. 1346 on 60
mg Raloxifene 3 year follow-up. - Teriparatide vs. Placebo (Eli Lilly Co.)
- 189 osteoporotic women on placebo vs. 186 taking
20 ?g and 183 taking 40 ?g rhPTH(1-34), 21 month
follow-up. - Alendronate vs. Placebo (Merck Co.)
- 155 osteoporotic women on Placebo vs. 164
Alendronate 3 year follow-up.
35Effects of Raloxifene60 mg Raloxifene vs.
Placebo at 3 year follow-up
p lt 0.05 vs. Placebo
Uusi-Rasi, K, et, al. 2006 Osteoporos. Int
36Effects of TeriparatideTeriparatide vs. Placebo
at 21 Month Follow-up
p lt 0.05 vs. Placebo
Uusi-Rasi, K, et al. 2005, Bone
37Effects of Alendronate Differences From Placebo
at 3 Year Follow-up
plt0.05 vs. Placebo
(Poster 211 International Society of Clinical
Densitometry 2006)
38Lessons
- BMD decline with age is partially due to
expansion of outer dimensions - Expansion serves to preserve bending strength but
with less bone material - An expanded bone may be structurally unstable
when subjected to unaccustomed loading forces - This may be the geometric explanation for why low
density bones are fragile
39Lessons II
- Clinically proven treatments produce improvements
in femur geometric strength and cortical stability
40Measuring bone geometry from DXA
- Specialized software is required
- Will be commercially available from DXA
manufacturers - Better scan quality control is needed especially
in subject positioning - Better quality scan images will be needed
- Also becoming available on newer scanners
- Certification of operators?
41Scans Unsuitable for HSA
42Becks Hip Structure Analysis
- Over 30 Published Studies
- Most Recognized 2D Method of Structural Analysis
- Integrated into Hologic Software
- All standard HSA parameters available in Hologic
MS-Access database.
43(No Transcript)
44Hologic HSA refinements
- BMD matches Hologic BMD with all of Hologics
proprietary corrections - ROIs are in mms, not points or line sizes,
allows for standardization between patients and
to future scan types and models - Magnification effects more accurately removed
when the new Auto Hip Positioning option is used
45Auto Hip Positioning
- User positions laser on greater trochanter
- Scanner automatically moves down and then scans
across the shaft - While scanning across the shaft, triangulates to
measure height of femur off the table - Finds shaft for perfect positioning of femur and
automatically starts scan