Date post: | 28-Oct-2018 |
Category: |
Documents |
Upload: | truongmien |
View: | 216 times |
Download: | 0 times |
MRI: What Can It Deliver
Claude B. Sirlin, MD
Liver Imaging Group, Dept. of Radiology
University of California, San Diego
2nd International
NASH Biomarkers WorkshopWashington, DC May 5, 2017
Email me for PPT:[email protected]
Industry grant support
Siemens, GE, Guerbet, Bayer, Philips
Consulting and service agreements
Alexion, AstraZeneca, Bioclinica, BMS, Bracco, Fibrogen,
Galmed, Genentech, Genzyme, Gilead, Icon, Intercept, Isis,
Janssen, NuSirt, Perspectum, Pfizer, Profil, Sanofi, Shire,
Synageva, Tobira, Takeda, Virtual Scopics
Federal grant support
U01 DK061734, R01 DK106419, R01 DK088925, T32
EB005970, R01 DK088831
Disclosures
Today• Body composition
• Proton density fat fraction (PDFF)
• Corrected T1
• “Stiffness”
Tomorrow• Beyond “stiffness”
The future• Pie in the sky
• Intrinsic meaning
What Can MRI Deliver
< 30 min
Today• Body composition
•
•
•
•
•
•
What Can MRI Deliver
Body Composition
Magnus BorgaOlof Leinhard
MRI-based whole body composition profiling
Advanced MR Analytics (AMRA)
Feasible and Precise
Exam = 5-10 minutes
Exam = 5-10 minutes
Exam = 5-10 minutes
Manual analysis = 15 hours
Manual analysis = 15 hours
Manual analysis = 15 hours
Exam 1 Exam 2 Exam 3
Semi-automated analysis = minutes; coefficient of variation 1.5-3.6%
Body Composition
Weight loss BMI 41 kg/m2 BMI 33 kg/m2 BMI 30 kg/m2
VAT 5.7 L VAT 4.1 L VAT 3.1 L
SCAT 18.3 L SCAT 12.0 L SCAT 9.7 L
Longitudinal monitoring
Body Composition
Today•
• Proton density fat fraction (PDFF)
•
•
•
•
•
What Can MRI Deliver
PDFF
Proton Density Fat Fraction (PDFF)
S(TE) = [W + F rpei 2p DfpTE
p=1
P
å ]ei 2p yTEe-TE T2*
Hero HussainW Tom Dixon Takeshi YokooMark Bydder Michael Middleton Gavin HamiltonScott Reeder Huanzhou Yu Jean Brittain An Shimakawa
PDFF = proportion of all MRI visible protons from fat
Water Fat
PDFF
PDFF = proportion of all MRI visible protons from fat (triglyceride)
Steatosis Triglyceride
PDFF
≈wikipedia/Trimyristin-3D-vdW
Proton density fat fraction50%: ~ maximum PDFF in human liver
3%: metabolic syndrome
5.5%: 95th percentile in normals
6.5%: hepatic steatosis
0% 100%
Lidia Szczepaniak, PhD
Jennifer Rehm, MD
An Tang, PhD
PDFF
Two Basic Approaches
Magnitude data-based MRI Complex data-based MRI
40 40
373640 41
38 3938 39 4039
3637
3638
928
100% 100%
0% 0%
PDFF
Magnitude data-based MRI Complex data-based MRI
40 40
373640 41
38 3938 39 4039
3637
3638
100%
0%
Two Basic Approaches100%
0%
PDFF
1.5TSiemens
3TGE
Geri Kang et al. JMRI 2011 Children & Adults with NAFLD
Liver proton density fat fraction is “portable”PDFF
Baseline T = 2 weeks T = 1.5 months T = 3.5 months T = 6.5 months
20.5% 17.2% 8.1% 6.5% 6.3%
Liver proton density fat fraction can monitor change longitudinally
100%
0%
50%
PDFF
Liver proton density fat fraction can be used as an endpoint in phase 2 clinical trials
RohitLoomba
PDFF
Today•
•
• Corrected T1
•
•
•
•
What Can MRI Deliver
cT1
Corrected T1
Rajarshi Banerjee Stefan Neubauer Matthew Robson
cT1 1055 ms
cT1 906 ms
cT1 869 ms
cT1 733 ms
cT1
Corrected T1
cT1 1055 ms
cT1 906 ms
cT1 869 ms
cT1 733 ms
Ishak6
Ishak3
Ishak2
Ishak0
600 ms 1200 ms≥ 800 ms
cT1
Corrected T1
Mixed population (viral hepatitis n=31, FLD n=31, other n=17)
Detection of any diseaseCutoff cT1 ≥ 800 msAUROC 0·94sensitivity 86% specificity 93%
Journal Hepatology Volume 60(1): 69-77 (2014)
cT1
T1corrections
cT1
cT1
1200 ms
600 ms
Free Water
Edema
Watery fibrosis
Bound Water
Collagen
Glycogen
cT1
Theory
Question: can cT1 separate the opposing effects of free and bound water?
Today•
•
•
• “Stiffness”
•
•
•
What Can MRI Deliver
Stiffness
“Stiffness”
Phil RossmanDick Ehman Raja Muthupillai
29 September 1995
Stiffness Ability to resist deformation from external pressure
Slide concept adapted from www.physicstutorials.org
Pressure Pressure
Stiff
Pressure Pressure
Not Stiff
Stiffness
How do you measure stiffness without palpation?
Sound (waves) travels faster through solid ground (hard) than through air (soft)
http://www.corbisimages.com/stock-photo/rights-managed/DEC452-14/indian-scout-putting-his-ear-to-ground
Stiffness
Shear Wave Propagation in Liver
Stiffness
MR Elastography Diagnoses Advanced Fibrosis
“Stiffness” cutoff: 3.63 kPaSensitivity 0.86Specificity 0.91
Loomba et al 2014
Dick EhmanRohit Loomba David Brenner
Anthony Gamst Jonathan HookerAUC for diagnosis of advanced fibrosis
0.924
Stiffness
Singh et al 2015
MR Elastography Not Sensitive to Early Fibrosis
Meta-Analysis
• 12 Studies• 697 patients
Stiffness
Stiffness
NASH: Advanced Fibrosis
Collagen Stiffness
NASH: Early Fibrosis
Conventional MRE (“stiffness”) cannot differentiate mild fibrosis from inflammation
•
•
•
•
Tomorrow• Beyond “stiffness”
•
•
What Can MRI Deliver
Beyond Stiffness
Beyond Stiffness
Beyond Stiffness
Beyond Stiffness Stiffness has elastic and viscous components
Slide concept adapted from www.physicstutorials.org
Pressure Pressure
Elastic
Returns to original size and shape
Pressure Pressure
Viscous
Does not return to original size and shape
Pressure released Pressure released
Beyond Stiffness
Elastic Viscous
bouncy ball silly putty
Beyond Stiffness Stiffness has elastic and viscous components
Beyond Stiffness
Elastic Viscous
Fibrosis Inflammation
Beyond Stiffness Stiffness has elastic and viscous components
Beyond Stiffness 3D MR elastography analyzes wave motion in all 3 planes
Beyond Stiffness
Shea
r St
iffn
ess
(kP
a)
0
5
2
3
4
1
X Y Z |G*|
Slide courtesy of Meng Yin, PhD
Beyond Stiffness 3D MR elastography analyzes wave motion in all 3 planes
Beyond Stiffness
Dick Ehman Meng Yin
stiffness. As shown in
Figures 5 and 6, G’ typically dominates G”, and
the normal damping ratio is
small. 3D-MRE also can be
processed to filter out the shear waves, leaving only
the longitudinal data.
Inversion of this data can generate maps of the bulk
modulus or “volumetric
strain” (ev), a marker of compressive deformation.
ev has been shown to correlate with portal pressure78, and so may
provide insights into
NAFLD pathophysiology,
which this application will explore. In summary, 3D-MRE can generate
multiparametric maps of mechanical properties (|G*|, G’, G”, ζ, ev), as shown in two examples of obese patients with NAFLD (Figure 6). The relevant
technology is operational at UW and UCSD.
Multiparametric 3D-MRE Detects NASH in Animal Models. In mouse
models of NASH79, one of our co-investigators, Dr. Yin, showed that hepatic
inflammation correlates with the loss modulus G”, while fibrosis correlates with the storage modulus G’25. Similarly, she observed that G” and the damping
ratio are elevated in early NASH. This permitted detection of NASH prior to
fibrosis development (Figure 7). Since the storage modulus, G’, is the dominant component of G*, the shear stiffness |G*| is relatively insensitive to
inflammation, which may explain the limited efficacy of 2D-MRE for detecting
early NASH. By comparison, multiparametric 3D-MRE quantifies inflammation-
sensitive parameters (G”, damping ratio), potentially permitting early NASH detection. This will be examined in our proposed study.
Complex Modulus Depends on Frequency: Optimization & Opportunity.
i) Optimization of single best frequency to quantify complex modulus: Mechanical frequency is a key determinant of the quality and analyzability of
the wave field. Maximizing wave quality is particularly important in obesity,
because large body habitus may hinder wave propagation. Previous work by our group has shown that frequencies below 30Hz generate shear waves prone to interference effects and are
unreliable (unpublished), while frequencies above 75 Hz result in rapid80,81. Thus, the optimal frequency for
MRE is thought to be in the 30-75 Hz range but the single optimal frequency is unknown, a knowledge gap that
Phase 1 will address.
ii) Opportunity to examine the dispersion of viscoelastic parameters of liver: Frequency also affects the
magnitude of MRE mechanical parameters, a phenomenon known as “dispersion”. For most tissues, |G*|, G’,
and G” increase as the shear wave frequency increases. Importantly, different tissues and pathologic conditions (normal liver, inflammation, fibrosis) have different dispersion characteristics. Multifrequency MRE
enables dispersion characterization, potentially assisting in the discrimination between inflammation, fibrosis,
and other processes. Numerous physical models have been proposed to model dispersion based on underlying assumptions of viscoelastic properties of tissue. An incomplete list includes Voigt, Maxwell, Zener,
Jeffreys, fractional Zener, springpot-1, springpot-2, springpot-3, and fractional derivative models82-84. These
models quantify numerous viscoelastic parameters some of which correlate with the fibrosis, primarily in viral
hepatitis11
. To date, there are no published data using multifrequency dispersion to evaluate obesity-associated NAFLD in humans.
A problem with measuring viscoelastic dispersion is that they typically require MRE acquisition at three or
more frequencies, limiting clinical practicality. Dr. Yin recently showed, however, that multifrequency dispersion
Figure 6: 3D-MRE can quantify multiparametric mechanical properties of liver in obesity.
Magnitude images and maps of complex shear modulus (G*=G’+iG”), damping ratio (ζ
=G”/(2G’)), and volumetric strain (ev), acquired at 60Hz are shown. Patient 1 (top row,
23yo F, BMI=44kg/m2) has NASH but no fibrosis at biopsy. Liver “stiffness” (|G*|) is
normal. Damping ratio (ζ) is elevated (based on preliminary data, normal ζ≈0.05-0.08).
Patient 2 (bottom row, 28yo F, BMI=43 kg/m2) has NASH with stage 4 fibrosis. Liver
“stiffness” (|G*|) and damping ratio are both elevated. As illustrated in Patient 1, damping
ratio shows promise for detecting NASH even if overall stiffness is normal.
Figure 7: Mouse model of NASH. Damping ratio detects NASH prior to fibrosis development. Liver stiffness (|G*|) detects fibrosis.
Data obtained in mice fed fast
food diet for 1, 24, and 48 weeks
NASH Fibrosis No liver injury
Da
mp
ing
R
ati
o
1
2
3
Sti
ffn
es
s
(kP
a)
1
2
2.5
1.5
Inflammation?Fibrosis
Beyond Stiffness 3D MR elastography measures multiple tissue parameters
Beyond Stiffness
•
•
•
•
•
The future• Pie in the sky
•
What Can MRI Deliver
Future
https://media.licdn.com
Pie in the sky• Automation
• Blood flow
• Liver function
• Fat composition
• Glycogen content
• Fibrogenesis
• Fibrolysis
• Portal pressure
• Pancreatic beta cell function
• Metabolic pathways
• Stellate cell activation
• Mitochondrial function
•
•
•
•
•
The future•
• Intrinsic meaning
What Can MRI Deliver
Future
Intrinsic meaning50
0
%
PDFF 25.0%
Chemical Shift MRI
8
0
kPa
Shear stiffness 3.23 kPa
MRE
LIF
Liver Inflammation and Fibrosis Score 2.7
Liver MultiScan
Steatosis Fibrosis
https://upload.wikimedia.org/wikipedia/commons/8/8a/Steatohepatitis_high_mag.jpg
Inflammation and Fibrosis
PDFF Shear stiffness Corrected T1
•MRI measurements no intrinsic meaning
•Must be translated into a histology score
Current Paradigm
Future
Intrinsic meaning50
0
%
PDFF 25.0%
Chemical Shift MRI
8
0
kPa
Shear stiffness 3.23 kPa
MRE
LIF
Liver Inflammation and Fibrosis Score 2.7
Liver MultiScan
PDFF Shear stiffness Corrected T1
•MRI measurements acquire intrinsic meaning
•Do not have to be translated into a histology score
FutureParadigm
[Triglyceride] within
MRI visible protons
Elastic and viscous components of shear stiffness
Water content within
MRI visible protons
Outcomes
Future
Intrinsic meaning
StiffnessElasticity
PDFFcT1
Beyond stiffness
Tetri Grid
Future
Intrinsic meaning
StiffnessElasticity
PDFFcT1
Beyond stiffness
Today• Body composition
• Proton density fat fraction (PDFF)
• Corrected T1
• “Stiffness”
Tomorrow• Beyond “stiffness”
The future• Pie in the sky
• Intrinsic meaning
What Can MRI Deliver