1

Clinical Assessment of Bone Strength and Fracture Risk: Beyond BMD

Mary L. Bouxsein, PhD Center for Advanced Orthopedic Studies, BIDMC

Department of Orthopedic Surgery, HMS MIT-Harvard Health Sciences and Technology Program

Disclosures

•  Advisory board: Merck, Eli Lilly •  Consultant: Agnovos •  Research funding: Merck, Amgen

2

Structural failure of the skeleton

Design of a Structure

•  Consider what loads it must sustain

•  Design options –  Overall geometry –  Building materials –  Architectural details

3

Determinants of Whole Bone Strength

Morphology size (mass) shape (distribution of mass) porosity microarchitecture

Properties of Bone Matrix mineralization collagen microdamage non-collagenous proteins ..and more…

Mechanisms underlying bone strength at multiple length scales

collagen : mineral interactions

Zimmerman et al, PNAS 2011 Zimmerman et al, PNAS 2011

4

Standard clinical assessment of bone strength

Areal BMD by DXA •  Bone mineral / projected area (g/cm2) •  Reflects (indirectly)

–  Bone size –  Mineralization

•  Moderate to strong correlation with whole bone strength (r2 = 50 - 90%)

•  Strong predictor of fracture risk in untreated women (Marshall et al, 1996)

BMD explains > 70% of whole bone strength in ex vivo human cadaver studies

0

2000

4000

6000

8000

10000

0 0.2 0.4 0.6 0.8 1 1.2 Femoral Neck BMD (g/cm2)

Fem

oral

Str

engt

h (N

)

r2 = 0.71

Bouxsein © 1999

Moro et al, CTI, 1995

Ver

tebr

al (

L2)

Fai

lure

Loa

d (lb

s)

0 . 2 0 . 4 0 . 6 . 8 1 0

1000

2000

3000

4000

5000

6000

. . . 0 . Spine BMD (g/cm2)

Proximal Femur (sideways fall) Vertebral body (L2)

(compression + forward flexion)

r2 = 0.79 Does not distinguish several attributes of whole bone strength

–  3D geometry –  Microarchitecture –  Intrinsic properties of bone matrix

5

0

5

10

15

20

25

30

35

40

45

50

BMD T-scores

0

50

100

150

200

250

300

350

400

450

Many fractures occur in those who have BMD t-scores better than -2.5

Frac

ture

s pe

r 100

0 pe

rson

yea

rs

# W

omen

with

Fra

ctur

es

Fracture rate # Women with Fractures

E. Siris, et al, NORA study, > 200,000 women in US

BMD has limitations in clinical use

•  Less than half of patients who fracture have osteoporosis by BMD testing (ie t-scores > -2.5*) –  Only half of elderly women with incident hip frx had

BMD in osteoporotic range at baseline

•  Change in BMD underestimates anti-fracture efficacy of drugs

•  Measurements are subject to artifacts –  Obesity, vascular calcification, OA

Schuit et al, 2004; 2006; Wainwright et al , JCEM 2005; Cummings et al 2002; Sarkar et al, 2002; Watts 2004, 2005; Bruyere et al, 2007a,b; Austin et al, 2011

6

Men have larger but less dense vertebrae than women matched for aBMD

Men Women Diff (%) Cross-sectional area (cm2) 12.39 10.33 +20%* Integral vBMD (g/cm3) 0.156 0.215 -8%* Trabecular vBMD 0.137 0.150 -9%* Compressive strength (N) 4425 3003 +10%*

981 pairs Spine aBMD (± 1%) Age (± 1 yr)

Bone Strength

SIZE & SHAPE how much?

how is it arranged?

MATRIX PROPERTIES mineralization collagen traits

BONE REMODELING formation / resorption

Osteoporosis Drugs, Diet, Exercise, Diseases, ….

Non-Invasive Imaging

Bone Turnover Markers

?

7

DXA-based methods

QCT, HR-QCT

HR-pQCT

QCT-based Finite Element Analysis

Reference point indentation

DXA-based methods

QCT, HR-QCT

HR-pQCT

Finite Element Analysis

Reference point indentation

8

Hip Structure Analysis (HSA)

•  Uses standard 2D DXA scans to estimate femoral geometry

Key assumptions: •  Constant mineral density •  Neck, Shaft = circular •  Troch = elliptical •  % of cortical bone constant

–  Shaft = 100% –  Fem neck = 60% –  Troch = 70%

•  Highly correlated to femoral BMD

Trabecular bone score (TBS) (FDA-approved in 2012)

Silva et al JBMR 2014 (accepted)

↑↑TbN ↓TbSp ↑Conn D

9

TBS associated with fractures, weakly with BMD •  29,407 postmenopausal women; 1668 (5.6%) had major OP frx •  Weak correlation to BMD: r = 0.26-0.33

Hans et al JBMR 2011

Incident fracture

(per 1000 person-yr)

Silva et al JBMR 2014

Cross-sectional studies

TBS and Fracture

10

TBS enhances 10-yr prediction of fx risk •  33,352 women (40 - 100 yrs), 4.7 yrs of follow-up

–  1,754 deaths and 1,872 major osteoporotic fx

Leslie et al Osteop Int 2014

Hazard Ratio (95% CI)

Major OP Fx Death

TBS 1.36 (1.30-1.42) 1.33 (1.27-1.39)

Spine BMD 1.47 (1.40-1.55) 1.03 (0.98-1.08)

Lowest TBS (~ 10th percentile) 1.5 - 1.6 fold greater risk of major OP fracture

TBS enhances 10-yr prediction of fx risk

Leslie et al Osteop Int 2014

(FRAX)

11

TBS updates: towards clinical utility •  Age-specific reference values

–  US non-hispanic white women (Simonelli et al JCD 2014) –  Japanese women (Iki et al, Osteop Int 2015)

•  Association with fracture –  Men and women –  Women with osteopenia –  Diabetics

•  Utility in special patient populations Diabetes GIOP 1o HPTH Androgen deprivation CKD RA

Silva JBMR 2014; Maricic Curr Rheum Rep 2014; Leib et al Bone 2014; Leslie et al, JCEM 2013; Kim et al, JCEM 2015; Iki et al, JBMR 2014; Dhaliwal et al, Osteop Int 2014, Silva et al JCEM 2013

DXA-based methods

QCT, HR-QCT

HR-pQCT

Finite Element Analysis

Reference point indentation

12

QCT •  ~300-500 µm voxel size •  Trabecular & cortical bone

density, geometry •  Axial and appendicular

skeleton •  Associated with fracture •  Less sensitive to soft tissue

variability than DXA •  Novel applications

•  Limited micro-structure •  Few prospective fx studies •  No standard analysis •  Lack of reference data •  Radiation exposure?

Age-Related Changes in vBMD and Geometry by 3D-QCT

(368 women, 320 men, aged 20-97 yrs; Riggs et al JBMR 2004)

Lumbar spine Total area Trabecular vBMD Femoral neck Total area Trabecular vBMD Cortical vBMD

Women F vs M

(P-value) Men % Change, ages 20-90 yrs

For age regressions: *P<0.05, **P<0.005

+14** - 54**

+13** - 56** - 24**

+15** - 47**

+7* - 45** - 13**

0.253 <0.001 0.557 <0.001 <0.001

13

QCT-based femoral neck measures and hip fracture risk Hazard ratios per SD reduction

All models adjusted for age, BMI, race, clinic site 3347 men > 65 yrs, 42 incident hip fx

QCT HR per SD

↓↓ % cortical volume 3.4 (2.3–4.9) ↓ Fem Neck area (cm2) 1.6 (1.3–2.1) ↓ Trab vBMD (g/cm3) 1.6 (1.1–2.3)

Black, Bouxsein et al, JBMR 2008

QCT-based femoral neck measures and hip fracture risk Hazard ratios per SD reduction

All models adjusted for age, BMI, race, clinic site 3347 men > 65 yrs, 42 incident hip fx

QCT QCT + DXA HR per SD HR per SD

↓↓ % cortical volume 3.4 (2.3–4.9) 2.5 (1.6–3.9) ↓ Fem Neck area (cm2) 1.6 (1.3–2.1) 1.5 (1.1–2.0) ↓ Trab vBMD (g/cm3) 1.6 (1.1–2.3) 1.2 (0.8–1.9) ↓ Fem Neck aBMD (g/cm2) 2.1 (1.1-3.9)

Black, Bouxsein et al, JBMR 2008

14

QCT for Monitoring Treatment Response: Changes in Spine Bone Density by DXA and QCT: the PaTH trial

0

10

20

30

40

PTH PTH/ALN ALN

Mea

n C

han

ge

(%)

Black et al, NEJM, 2003

aBMD by DXAvBMD by QCT(Trabecular)

Femoral neck cortical thickness varies around neck circumference

15

QCT reveals asymmetry of cortical bone loss

at femoral neck

0.0 1.0 2.0 3.0 4.0

0.0 1.0 2.0 3.0 4.0

Cortical thickness (mm)

*

*

20 yr old 80 yr old

Regional cortical thickness in femoral neck may predict hip fracture better than femoral BMD

(AGES-REYKJAVIK study, 143 hip fx, 298 controls Multivariable regression, Hazard Ratio adjusted for age, ht, wt

Women (88 frx, 187 ctrl)

Men (55 frx, 111 ctrl)

Fem Neck Frx Sup-Ant CtTh 1.6 (1.0-2.8) 3.2 (1.8-5.7)

FN aBMD 1.2 (0.7-2.0) 1.4 (0.8-2.4)

Troch Frx Sup-Ant CtTh 1.6 (0.9-2.8) 2.3 (1.1-4.6)

Inf-Post CtTh - 2.4 (1.3-4.5)

FN aBMD 1.5 (0.8-2.7) 1.5 (0.7-3.1)

Johannesdottir et al, Bone 2011

16

Cortical thickness mapping to identify local osteoporosis

Poole et al, PLoS ONE, 2011

Cortical deficits and fracture location in

acute hip fx

 QCT-based cortical thickness mapping Poole et al 2011, 2013; Treece et al 2010, 2013  124 controls  144 acute hip fx

–  53 trochanteric –  37 transcervical –  54 subcapital

Poole et al, ASBMR 2013

17

DXA-based methods

QCT, HR-QCT

HR-pQCT

Finite Element Analysis

Reference point indentation

82 µm voxel size Peripheral skeleton only < 3 µSv

HR-pQCT

18

Separation of cortical and trabecular compartments

HR-pQCT discriminates osteopenic women with and without history of fragility fracture

(age = 69 yrs, n=35 with prev frx, n=78 without fracture)

* p < 0.05 vs fracture free controls

Boutroy et al, JCEM (2005)

-12.4 -8.5

-5.6

12.8

25.6

-15 -10 -5 0 5

10 15 20 25 30

-0.3 0.4

-15 -10 -5 0 5

10 15 20 25 30

Frac

ture

vs

No

Frac

ture

, %

Diff

eren

ce

Femoral Neck

BV/TV TbN TbTh

TbSp TbSpSD

Spine

*

*

*

*

BMD HR-pQCT at the Radius

19

Osteopenic by DXA BMD (70 yr old woman)

Distal Tibia

Distal Radius

Worse bone architecture in premenopausal women with distal radius fracture

(40 premenopausal wrist fx, 80 age-matched controls)

DXA HR-pQCT

ΔFx

- C

ont (

%)

* P <0.05 * P < 0.05 UDR adj

* *

*

*

Wrist Frx

Control

Rozental, Bouxsein et al, JBJS (2013)

20

Race-related differences in microarchitecture

Pathophysiology of fragility in Type II Diabetes ?

Control Diabetes Diabetes + Frx

Burghardt et al, J Clin Endo Metab (2010)

T2DM have same or higher BMD, but markedly higher (+36 to 120%) cortical porosity

vs controls

21

T2DM with history of fracture have increased cortical porosity

Patsch et al, JBMR 2013

Con Con+Fx T2DM T2DM+Fx

DXA-based methods

QCT, HR-QCT

HR-pQCT

Finite Element Analysis

Reference point indentation

22

3D geometry

Crawford, et al, Bone 2003 QCT Density

Ulti

mat

e St

reng

th

material properties

Hip and spine FEA Approved by FDA in 2012

ON Diagnostics

Image courtesy of T. Keaveny

0

2000

4000

6000

8000

10000

Fem

ora

l Fai

lure

Lo

ad (

N)

0 2000 4000 6000 8000 10000 FEA-Predicted Femoral Strength (N)

Men Women

r2 = 0.78p < 0.001

Experimentally measured femoral strength vs FEA-predicted strength

(sideways fall, 73 human femora, aged 55 to 98 yrs) Bouxsein © 1999

23

Finite element analysis and hip fracture

QCT-based FEA vs femoral BMD for prediction of hip fracture

Fem Neck BMD

QCT-FEA strength

Older men (Mr OS)1 40 hip fx vs 210 controls

4.4 (2.4-9.1) 6.5 (2.3-18.3)

Older men (AGES)2 Older women (AGES) 171 hip fx vs 877 control

3.7 (2.5-5.6) 2.7 (1.9-3.9)

3.5 (2.3-5.3) 4.2 (2.6-6.9)

1Orwoll et al, JBMR 2009; 2Kopperdahl et al, JBMR 2014

* Adjusted for age, race

Hazard Ratio (95% CI)*

24

CT procedures amenable to bone strength evaluation - CT colonography, CT Enterography

+

Osteoporosis screening in IBD patients undergoing contrast-enhanced CT Enterography

R2 = 0.84 Women

Y Men

Weber et al, Am J Gastroenterology, 2014

25

DXA-based methods

QCT, HR-QCT

HR-pQCT

Finite Element Analysis

Reference point indentation

Bone Strength

SIZE & SHAPE how much?

how is it arranged?

MATRIX PROPERTIES mineralization collagen traits

BONE REMODELING formation / resorption

Osteoporosis Drugs, Diet, Exercise, Diseases, ….

?

26

How do we assess bone material and tissue level mechanical properties?

Estimated • elastic modulus • ult. strength • toughness

• elastic modulus • ult. strength • toughness

• elastic modulus • hardness

Assessing bone tissue level biomechanical properties with reference probe indentation

Hansma et al 2008, Hansma et al 2009, Diez-Perez et al JBMR 2010

Tota

l Ind

enta

tion

Dis

tanc

e (μ

m)

Age (years)

Hip Fracture Control

2.5 µm

27

Osteoprobe •  Hand-held, portable •  Single impact indentation •  Bone material strength index (BMSi)

Bridges et al, Rev Sci Instr 2012

In vivo reference point indentation

Randall et al, J Med Devices 2012

28

Indentation properties worse in women with T2DM

  30 postmenopausal women with T2DM (65.5±8.1 yrs) for > 10 yrs   30 age-matched, non-diabetic   No difference in bone microarchitecture b/w groups (HR-pQCT)   T2DM had 10.5% lower bone material strength (p<0.001)

Farr et al, JBMR 2014

Atypical femoral fractures: issue with cortical bone quality ?

 NIH-funded study to examine utility of RPI  Postmenopausal women:

–  Treatment naïve –  Long-term bisphosphonate users –  Atypical femoral fracture

29

Force applied to bone magnitude, direction

Bone strength

Risk of fall

Neuromuscular function Medications Environmental hazards Time spent at risk

Fall direction, height Protective responses Energy absorption Lifting activity

Fracture? Bone mass Bone geom + µ-arch Bone matrix prop’s

Peak bone mass Rate of bone loss

30

Biomechanics of Vertebral Fractures

•  Difficult to study –  Definition is controversial –  Many do not come to clinical attention –  Slow vs. acute onset –  The event that causes the fracture is often

unknown

•  Poor understanding of the relationship between spinal loading and vertebral fragility

Occurrence of vertebral fractures varies along the spine

T4 T6 T8 T10 T12 L2 L4 T4 T6 T8 T10 T12 L2 L4

31

FFRACTURE?

Vertebral strength Loads applied to the vertebra

Activity Spinal Curvature

Alex Bruno – HST PhD candidate

Estimating Loads on the Lumbar Spine

32

Standing 51 Rise from chair 173

Stand, hold 8 kg, 230 arms extended

Stand, flex trunk 30o, 146 arms extended

Lift 15 kg from floor 319 for a 162 cm, 57 kg woman

Activity Load (% BW)

Predicted Loads on Lumbar Spine for Activities of Daily Living

Vertebral loading predicts fracture location

0

100

200

300

400

500

600

700

800

900

T1 T3 T5 T7 T9 T11 L1 L3 L5

Vert

ebra

l Com

pres

sive

Loa

d (N

)

0

500

1000

1500

2000

2500

3000

T1 T3 T5 T7 T9 T11 L1 L3 L5

Vert

ebra

l Com

pres

sive

Loa

ding

(N)

T11 L1

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

T6 T8 T10 T12 L2 L4

Fact

or-o

f-ris

k

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

T6 T8 T10 T12 L2 L4

Fact

or-o

f-Ris

k

TK = 0° TK = 26° TK = 50° TK = 74°

T12

33

Non-invasive assessment of bone strength: where are we today?

•  A few techniques have recent FDA approval –  TBS may add to BMD predictions of fracture –  QCT-based FEA may expand number of individuals diagnosed

•  Techniques are well suited for clinical research and clinical trials –  Pathophysiology & differentiate mechanism of action

•  Not yet clear how to use in routine clinical practice –  QCT & FEA: No clear advantage over BMD for fx prediction –  Indentation -- early stage & many questions remain

•  Examining non-BMD aspects of fracture risk (ie, loading) provides important insights into mechanisms

Happy ΠΠ-day! 3.14.15 9:26:53

NIH AR053986Funding

Funding support: •  NIH R01 AR053986, R01 AR/AG041398, T32

AG023480, NHLBI Framingham Heart Study, and research grants from Amgen and Merck & Co.