Christine Baudelet et al 2004 Phys. Med. Biol. 49 3389-3411
Physiological noise in murine solid tumours using T2*-weighted
gradient-echo imaging: a marker of tumour acute hypoxia?
Christine Baudelet1,2, Réginald Ansiaux1,
Bénédicte F Jordan1,2, Xavier Havaux3, Benoit Macq4 and
Bernard Gallez1,2,5
1 Laboratory of Biomedical Magnetic Resonance, Université
Catholique de Louvain, Brussels, Belgium
2 Laboratory of Medicinal Chemistry and Radiopharmacy,
Université Catholique de Louvain, Brussels, Belgium
3 Cardiovascular Pathology Unit, Université Catholique de
Louvain, Brussels, Belgium
4 Communications and Remote Sensing Unit, Université Catholique
de Louvain, Louvain-La-Neuve, Belgium
5 Address for correspondence: CMFA/REMA Units, Université
catholique de Louvain, Avenue E. Mounier 73.40, B-1200 Brussels,
Belgium.
T2*-weighted gradient-echo magnetic resonance imaging (T2*-weighted GRE
MRI) was used to investigate spontaneous fluctuations in tumour
vasculature non-invasively. FSa fibrosarcomas, implanted
intramuscularly (i.m.) in the legs of mice, were imaged at 4.7 T, over
a 30 min or 1 h sampling period. On a voxel-by-voxel basis, time
courses of signal intensity were analysed using a power spectrum
density (PSD) analysis to isolate voxels for which signal changes did
not originate from Gaussian white noise or linear drift. Under baseline
conditions, the tumours exhibited spontaneous signal fluctuations
showing spatial and temporal heterogeneity over the tumour.
Statistically significant fluctuations occurred at frequencies ranging
from 1 cycle/3 min to 1 cycle/h. The fluctuations were independent of
the scanner instabilities. Two categories of signal fluctuations were
reported: (i) true fluctuations (TFV), i.e., sequential signal increase
and decrease, and (ii) profound drop in signal intensity with no
apparent signal recovery (SDV). No temporal correlation between tumour
and contralateral muscle fluctuations was observed. Furthermore,
treatments aimed at decreasing perfusion-limited hypoxia, such as
carbogen combined with nicotinamide and flunarizine, decreased the
incidence of tumour T2*-weighted GRE fluctuations. We also tracked
dynamic changes in T2* using multiple GRE imaging. Fluctuations of T2*
were observed; however, fluctuation maps using PSD analysis could not
be generated reliably. An echo-time dependency of the signal
fluctuations was observed, which is typical to physiological noise.
Finally, at the end of T2*-weighted GRE MRI acquisition, a dynamic
contrast-enhanced MRI was performed to characterize the
microenvironment in which tumour signal fluctuations occurred in terms
of vessel functionality, vascularity and microvascular permeability.
Our data showed that TFV were predominantly located in regions with
functional vessels, whereas SDV occurred in regions with no contrast
enhancement as the result of vessel functional impairment. Furthermore,
transient fluctuations appeared to occur preferentially in
neoangiogenic hyperpermeable vessels. The present study suggests that
spontaneous T2*-weighted GRE fluctuations are very likely to be related
to the spontaneous fluctuations in blood flow and oxygenation
associated with the pathophysiology of acute hypoxia in tumours. The
disadvantage of the T2*-weighted GRE MRI technique is the complexity of
signal interpretation with regard to pO2 changes. Compared to
established techniques such as intravital microscopy or histological
assessments, the major advantage of the MRI technique lies in its
capacity to provide simultaneously both temporal and detailed spatial
information on spontaneous fluctuations throughout the tumour.