Synopsis
Dynamic susceptibility
contrast (DSC-MRI) MRI is routinely used for brain tumor imaging and has shown
promise as an early biomarker for treatment response. Conventional DSC-MRI is susceptible
to contrast agent leakage effects, reducing the reliability of the
resulting blood volume maps. The use of a simplified spin and gradient echo (SAGE) sequence,
combined with robust processing strategies for correction of leakage effects, could
facilitate more rapid clinical translation and adoption of DSC-MRI for brain
tumor imaging. Taken together, the simplified SAGE approach and subsequence leakage correction provides a clinically feasible strategy for the simultaneous assessment tumor
perfusion, permeability and cellularity. Purpose
Perfusion imaging provides
a more complete assessment of brain tumor hemodynamics and may prove
particularly useful in the diagnosis and treatment of brain tumors. A combined
spin- and gradient-echo (SAGE) approach, consisting of two gradient echoes (GE),
two asymmetric spin echoes and one true spin echo (SE), was proposed as a means
to assess perfusion (including the total and microvascular blood flow and blood
volume), vessel size, and permeability (through the extraction of DR1 DCE-MRI
data).
1
A key advantage of SAGE is that contrast agent (CA)-induced T
1
leakage effects can be removed through the use of multiple echoes, thereby
improving the reliability of tumor blood volume measurements.
2,3 SAGE also enables the estimation of the
transverse relaxivity at tracer equilibrium (TRATE), a new DSC-MRI parameter
that reflects cellular features (cell density, polydispersity, and cell size).
4 We recently
proposed and validated in pre-clinical tumor models a simplified SAGE approach
that employs a combined dual GE and SE pulse sequence and an analytic solution
for computing T
1-insensitive ΔR
2* and ΔR
2 that
significantly increased the computational efficiency.
5 The combination of
the simplified SAGE data with our recently validated T
2*
leakage correction method
6 provides a
clinically efficient and robust approach for the derivation of leakage
corrected blood volume maps in brain tumors. Taken together, the simplified
SAGE approach provides a clinically feasible strategy for the simultaneous
assessment tumor perfusion, permeability and cellularity. The goals of this
study are to further validate simplified SAGE in glioma and brain metastases
patients and to demonstrate its utility for assessing Bevacizumab treatment
response.
Methods
Data were acquired at 3T
(Achieva, Philips Healthcare) in high-grade glioma patients before and after
Bevacizumab treatment (n = 4) and in brain metastases patients (n = 7). DSC-MRI
data were acquired using a SAGE-EPI sequence (TR = 1.8s, TEs = 8.8/26/55/72/88ms,
SENSE = 2, voxel size = 2.5 x 2.5 x 5.0mm
3, 15 slices, 7.5min
duration) before, during, and after administration of 0.1 mmol/kg Gd-DTPA. The T
1-corrected
∆R
2 and ∆R
2* were obtained analytically from the
GE and SE signals using the sSAGE equations
5 and were
subsequently corrected by subtracting the ΔR1-derived veCe(t),
scaled by the equilibrium transverse relaxivity, from the T
1-corrected
ΔR
2 and ∆R
2*.
6 The corrected DSC
perfusion data are compared to uncorrected DSC data using the 2
nd GE
and the SE. The derived hemodynamic parameters include GE and SE CBV, mean
vessel diameter (mVD), TRATE,
Ktrans,
and
ve.
Results
Figure 1 shows example GE
(top) and SE (bottom) DSC data in a high-grade glioma prior to treatment with
Bevacizumab. In tumor, T
1-shortening effects due to Gd-DTPA
extravasation manifest as lower post-bolus ∆R
2* and ∆R
2
for the uncorrected single echo data. The SAGE and sSAGE curves, both corrected
for T
1 leakage effects, do not exhibit reduced post-bolus ∆R
2*
and are in close agreement. Further correction for T
2 and T
2*
leakage effects yielded similar curves for sSAGE and SAGE ∆R
2*
and ∆R
2. Conventional DSC-MRI provides GE CBV maps that are sensitive
to T
1-leakage effects, manifest here as a substantially reduced
tumor CBV (Figure 2). The sSAGE and SAGE maps are corrected for T
1,
T
2, and T
2* leakage effects and are in close
agreement. The TRATE maps are sensitive to an array of cellular features and
may provide unique information regarding the tumor cellular environment. This
method also provides ∆R
1 curves that can be used in
standard DCE pharmacokinetic models to provide Ktrans maps.
Conclusions
The simplified SAGE
technique leverages multiple echoes to provide T1-insensitive GE and
SE hemodynamic parameters, combined with a simple method to correct for
remaining T2 and T2* leakage effects. This
method is in close agreement with the more time-consuming conventional SAGE
method. The use of this multi-echo sSAGE approach, combined with robust
processing strategies for correction of leakage effects, could facilitate more
rapid clinical translation and adoption of DSC-MRI for brain tumor imaging.
DSC-MRI with the sSAGE approach provides a wealth of information about tumor
vascularity, vessel size and permeability, and cellular characteristics, which
has important implications for brain tumor patient management.
Acknowledgements
We would like to thank Dr. Paul Moots, MD for his collaboration
and our subjects for their participation.
This work was supported by NCI P30CA68485, 1R01CA158079,
and the Vanderbilt University VICC Young Ambassadors. References
1. Schmiedeskamp H, et al. Magn Reson Med (2012) 68(1):30.
2. Schmiedeskamp H, et al. J
Cereb Blood Flow Metab (2013) 33(5):732.
3. Stokes AM, et al. Magnetic Resonance Imaging
(2014) 32(10):1181.
4. Semmineh NB, et al. Magn Reson Med (2015) 74(3):772-84.
5. Stokes AM, et al. Magn Reson Med (2015) doi: 10.1002/mrm.25591 [Epub ahead of print].
6. Stokes AM, et al. Magn
Reson Med (2015) doi: 10.1002/mrm.25906. [Epub ahead of print]