Download Breast Cancer Trends 1973-1999

Spectroscopic Window on
Tumor Metabolism
Michael Garwood, Ph.D.
Univ. of Minnesota
Role of MRS in the Clinical Management of Cancer
•
Diagnosis:
 guide biopsy
 avoid unnecessary/risky biopsies
 ascertain aggressiveness/stage/prognosis
•
Treatment:
 guide choice of treatment
 identify non-responders early
→ alter treatment regime
 tool for follow up
High Res 1H MRS of Cells
Non-Malignant cells
Malignant cells
extract
in vitro
Ackerstaff et al., J Cell Biochem 2003
GPC → PCho switch
Aboagye et al., Cancer Res 1999
Choline-containing compounds
H H
C H
+
R-CH2-CH2-N -C H
C H
H H H
-
-
H
In vivo 1H MRS of breast cancer
First reported studies:
Roebuck et al, Radiol 1998; Gribbestad et al, JMRI 1998
1H
MRI
MRS
suppressed
water
invasive ductal
carcinoma
lipid
Choline
compounds
(tCho)
lipids
Frequency (ppm)
CMRR 4 Tesla
Infiltrating
ductal
carcinoma
P < 0.0008
Benign
Focal
Fibrosis
Jacobs MA, Barker PB, et al. Proton magnetic resonance spectroscopic imaging of human
breast cancer: a preliminary study. J Magn. Reson Imaging. 2004 Jan;19(1):68-75
Membrane Choline Phospholipid Metabolism
Lysophosphatidic
acid
Adapted from Aboagye EO, Bhujwalla ZM. Cancer Res 59:80-84 1999
Mechanisms of increased PC in cancer:
• Increased expression and activity of choline
kinase [Ramirez de Molina et al., Oncogene 2002]
• Higher rate of choline transport [Katz-Brull &
Degani, AntiCancer Res. 1996]
• Increased PLD activity [Noh et al., Cancer Lett.
2000]
• Increased PLA2 activity [Guthridge et al., Cancer
Lett. 1994]
Glioblastoma Multiforme (High Grade Tumor)
FLAIR
Cho
Right
T1
Cho
Cr
NAA
NAA
Lac
Left
PPM
4.0
3.0
2.0
1.0
slide courtesy of Peter Barker, Johns Hopkins U
Prostate Cancer
Cheng LL, FEBS Lett. 2001
Normal human prostate
Tumor-bearing prostate
MRI/MRSI Targeted, TRUS-Guided Biopsies
courtesy of J.Kurhanewicz, UCSF
MR targeted TRUS guided biopsy positive
5
PSA - 12 ng/ml
Two prior negative biopsies
The accuracy of cancer detection of
MRI/MRSI targeted biopsy in men with
prior negative biopsy ≈80%.
(Yuen et al, J. Urol. 2004; Prando et al,
Radiology 2005)
5
5
5
5
The sensitivity of TRUS guided biopsy is
reduced in large prostates and when the
cancer is located in difficult locations such
as the apex or in the anterior or lateral
aspects of the prostate.
Journal Urology 2000, 164(2) 400-404
Chemical Shift: Minimized with higher BW pulses
Standard pulses
Broadband pulses
Spectrum
900
DF
DF
DF
Gradients
RF
1800
90
180
180
Center Frequency (-235 Hz)
%CS : DF / BWRF
X
Courtesy of: G. Metzger
Y
Z
13
OVS with over-prescription
Courtesy of: G. Metzger
14
Prostate Spectroscopy at 3T: Single Voxel
Echo Time, Coupling and SNR
TE = 260 ms
TE = 100 ms
Courtesy of: P. Choyke & G. Metzger
15
Case Study: Slice 5
Cho
Cre
Cit
Sp
Courtesy of: P. Choyke & G. Metzger
16
Quantification
• Metabolite ratios (eg, tCho/NAA, (tCho+Cr)/Cit)
• External reference (eg, phantom of known conc)
• Reference to tissue water signal
3 Tesla
Normal breast
MRI Devices 4-ch coil
3x3x3 cm voxel
LASER Localization
a)
TE Averaging (60-300ms
in 128 increments)
NEX=2
tCho
b)
6
5
4
3
2
1
0
Breast Anatomy
Lobules
Adipose
tissue
Fibroglandular
tissue
Stroma
Fat
Tavassoli, 1999
•
Anatomy varies
greatly
•
Tissues are
distributed
heterogeneously
 Intravoxel lipids are
inevitable
Netter, 1997
Internal Referencing with Water
• NOT assuming constant water concentration
• Assuming a two-compartment model (water & fat)
and all tCho is in the aqueous compartment


f gain fT1 fT2
AtCho
[tCho] 

Awater
f gain fT1 fT2
[tCho] expressed in molal
units (mmol tCho/kg water)
No assumptions about volume
or density


water
tCho
water
1


tCho MWwater
A  Time domain amplitude
f gain  receiver gain correction
fT1 fT2  relaxation correction
water ,tCho  # nuclei / molecule
MWwater  molecular wt
Bolan et al., MRM 2003
Spectral Fitting
Adapted TDFDFit (Slotboom et al., MRM 1998)
 Time-Domain Model: s(t )  A  exp(it  i  t   2t 2 )
 Minimize residuals in frequency-domain over
narrow (0.4 ppm) band
model
• Fit 3 peaks independently:
tCho, water, 1.3 ppm lipid
data
• Errors from Cramer-Rao
Minimum Variance Bound;
used for detection threshold
residual
6
4
ppm
2
0
Bolan et al., MRM 2003
Normal gland (Presumed)
[tCho] = 0.75 ± 0.07 mmol/kg
volume = 13.0 mL
lipid fraction = 3.5%
Invasive Ductal Carcinoma
[tCho] = 6.8 ± 0.1 mmol/kg
volume = 6.8 mL
lipid fraction = 8%
Atypical Hyperplasia
[[tCho] = 1.5 ± 0.8 mmol/kg
volume = 1.1 mL
lipid fraction = 15%
Bolan et al., MRM 2003
no Cho
invasive ductal carcinoma
6
5
4
3
2
1
0
-1
-2
Frequency (ppm)
Reason for false negative? Spurious lipid sideband peaks!
invasive ductal carcinoma
6
5
4
3
2
1
0
-1
-2
Frequency (ppm)
Sideband Artifacts
water
sidebands
sidebands
• Antisymmetric
side peaks
TE (ms)
57
• Amplitude >1%
• Caused by B0
oscillation
45
-500
-300
-100
Hz
Sidebands have coherent,
TE-dependent phase
100
300
-500
Averaging causes
destructive interference
Bolan et al., MRM 2002
Echo-time Averaging
tCho?
NEX=64
TE=45ms
Conventional
single TE
No
tCho
TE=45-196ms
64 increments
TE averaging
8
6
2
4
ppm
0
-2
Bolan et al., MRM 2002
In vivo 1H spectrum of a voxel containing mainly adipose tissue
Day 127 (AC x 4 followed by Taxotere x 3)
size = 3.0 x 2.7 x 3.0 cm3
[Cho] = 0.642 mmol/kg
[Cho] = 0.910 mmol/kg
[Cho] = 0 mmol/kg
Invasive Ductal Carcinoma
Precontrast
Postcontrast
Subtraction
4
H2O
Lipid
3
Lipid
Lipid
SI 2
tCho
1
0
0
1
2
3
time (min)
(sec)
4
All 4 readers maintained
their decision to biopsy
5
7
6
5
4
3 2
ppm
1
0
-1
[tCho] = 0 ± 1.73
Meisamy et al, Radiology 2005
Conclusions about MRS for breast cancer
diagnosis:
 Adding quantitative 1H MRS to breast MRI improves
sensitivity, specificity, and accuracy, over MRI alone
 Quantitative 1H MRS is particularly useful in cases where
lesion morphology and time-intensity curves are
indeterminate
Meisamy et al, Radiology 2005
Treatment Planning and Monitoring
MRSI for Radiation Treatment Planning of Brain Tumor
MRSI-based radiation dose
painting using the IMRT method
Cho/Cr
Grade
Dose painting
<1
0
≥1-2
1
5040
≥2-3
2
5940
≥3
3
7020
Thakur, Chang, Huang, Koutcher, Narayana
Memorial Sloan-Kettering Cancer Center
Models of tCho response
Jordan et al., NMR Biomed 2006
Al-Safar et al., Cancer Res 2006
cell
density
PCho
CK
Measured acute response to PX-478
(inhibits HIF1-alpha production) in mouse
xenografts of HT-29 (colon)
Measured acute response to MN58b
(inhibits CK) in mouse xenografts of MDAMB-231 (breast) and HT-29 (colon)
Methods: in vivo MRS at 4.7T, ex vivo
validation
Methods: in vivo MRS at 4.7T, ex vivo
validation
Results: tCho dropped significantly at 12
and 24 hrs
Results: tCho dropped significantly at
48hrs in both models
Treatment Monitoring
in Breast Cancer
Neoadjuvant chemotherapy (primary systemic therapy,
PST) is the preferred treatment for locally advanced
breast cancer (Fisher et al. J Clin Oncol 1997, 1998)
Advantages:
 Tumor shrinkage; possible breast conserving
procedures
 In vivo monitoring of chemo-sensitivity
(customize Tx  complete pathologic response)
4T Tx Monitoring in Breast Cancer:
Results to Date
Non-Responders
9
9
8
8
7
7
[tCho] (mmol/kg)
[tCho] (mmol/kg)
Responders
6
5
4
3
5
4
3
2
1
1
0
0
Day 1
• 9/10 Nonresponders had a
increase in [tCho]
at Day 1
6
2
Baseline
• 14/18 Responders
had a decrease in
[tCho] at Day 1
• Day 1 Rule:
82% accuracy in
28 subjects
Baseline
Day 1
Meisamy et al, Radiology 2004
Responder to AC
Pre PST
24 hrs AC X 1
AC X 4
[tCho] = 4.6
LD = 4.0 cm
[tCho] = 3.7
LD = 4.0 cm
[tCho] = 0.9
LD = 1.7 cm
Meisamy et al, Radiology 2004
Responder to AC, but not Taxol
Pre PST
24 hrs AC X 1
AC X 4
Taxol X 2
[tCho] = 4.6
LD = 4.0 cm
[tCho] = 3.7
LD = 4.0 cm
[tCho] = 0.9
LD = 1.7 cm
[tCho] = 4.1
LD = 1.7 cm
Meisamy et al, Radiology 2004
Therapeutic Selection and Monitoring
Baseline
1 year
5 years
courtesy of J.Kurhanewicz, UCSF
citrate choline
Metabolic Atrophy
Metabolic Atrophy
Is it possible to predict response from
baseline MRS data?
Treatment Prediction / Phenotyping
Non-Responders
9
8
8
7
7
[tCho] (mmol/kg)
[tCho] (mmol/kg)
Responders
9
6
6
5
5
4
4
3
3
2
2
1
1
0
0
Baseline
Baseline [tCho] was
higher in
responders than in
non-responders
(p=0.03)
Day 1
Baseline
Day 1
Inconsistent findings in brain MRS:
Tzika, Neuroradiology 2001 – responders had
lower tCho
Preul, Neurosurgery 2000 – no difference
Lazareff, J Neurooncol 1999 – no difference
Higher [tCho] @
baseline associated
with higher grade &
positive nodes
Can MRS identify
responders before
starting treatment?
Pretreatment PME/NTP ratio
Preliminary results with 31P MRSI
Pretreatment 31P spectrum from nodal
disease of a HNSCC patient who
experienced partial response
Pretreatment PME/NTP ratios from tumors;
complete responders were different from
incomplete response group P<0.001
A. Shukla-Dave, et. al. Acad Radiol, 9:688-694, 2002
31P MRS in Bone Sarcoma
Baseline spectrum
Zakian, et. al., Cancer Research 2003 Dec 15;63(24):9042-7
Baseline Energetics Predicts Outcome in
Bone Sarcoma
NTP/Pi predicts longer survival
Zakian, et. al., Cancer Research 2003 Dec 15;63(24):9042-7
Future:
• More studies correlating with pathology,
immunohistochemistry, and outcomes
• Further studies to assess reliability/reproducibility
• Results of multi-center trials
• Combine with other metrics (DCE-MRI, ADC,…)
→ multiparametric analyses
• 3T (and higher?)
IMAPS (1.5T)
Data example
A
Prostate spectroscopy at 1.5T
with endorectal coil
C
The axial T2-weighted image (A) is used for
matching voxel locations to histopathological
specimens (D). One of the spectral maps (B),
partially expanded in (E), reflects the quality of
the MRSI data throughout the slice. Deviations
in the (Cho + Cr)/Ci metabolite ratio map in (C)
largely correspond to the tumor location
indicated with the blue line in (D).
D
B
Cho+Cr
Ci
Cho Ci
Courtesy of T. Scheenen and Prof. A.
Heerschap, Radboud University Nijmegen
E
Medical Center, Dept. of Radiology
Figure 3. The axial T2-weighted image (A) is used for matching The IMAPS community
voxel locations to histopathological specimens (D). One of the
Slide courtesy of Michael Jacobs, JHU
Current Multiparametric (MRI/DTI/MRSI) Prostate Imaging Exam
Slide courtesy
J. Kurhanewicz
UCSF
MRSI (0.3 cc)
Decreased
Signal Intensity
on T2 weighted
Imaging
Reduced
water
diffusion
T2 weighted MRI
Healthy
Cancer
Citrate
Elevated
choline
Reduced
citrate
Reduced
polyamines
Lipid
Choline
Creatine
Creatine
Polyamines
Choline
Diffusion weighted MRI
ADC Map
PPM 3.0
2.5
2.0
3.0
2.5
2.0
3T MRSI vs 1.5T MRSI: Improved Detection of Residual Cancer
3T
1.5T
Choline
Creatine
0.16 cc
0.34 cc
Cho
Cho
Cho
Cho
Cho
Acknowledgements
Thanks for Sending Slides
Arend Heerschap
Jason Koutcher
John Kurhanewicz
Michael Jacobs
Peter Barker
Wei Huang
U of Minn Researchers
Patrick Bolan
Greg Metzger
Sina Meisamy
Adeka McIntosh
Curt Corum
Angela Styczynski
Nate Powell
Djaudat Idiyatullin
Jang-Yeon Park
Carl Snyder
James Boyum
Doug Yee
Michael Nelson
Tim Emory
Lenore Everson
Todd Tuttle
Evin Gulbahce
Tommy Vaughan
Funding Sources
National Institutes of Health (CA92004, RR08079)