Magnetic resonance in neuro-oncology: Achievements and challenges - Signal Processing in MRS with Biomedical Applications - page 270

MRS. When lipid peaks are seen, this often reflects necrosis. Identification
of specific lipids has heretofore required the high resolution of in vitro MRS
applied to biopsy specimens from brain tumors [259].
In vitro NMR shows that triglycerides are typical spectral features of ac
tively growing nonnecrotic tumors. Highgrade gliomas have prominent neo
vascularity and highesterified cholesteryl. No cholesteryl esters were detected
in healthy brain tissue or in lowgrade and benign tumors [299]. Highly ester
ified cholesteryl could be a biochemical marker of malignancy. The presence
of cholesteryl esters and triglycerides has been found [300] to be correlated
with the degree of vascular proliferation in highgrade brain tumors.
The overlapping components of total choline within the spectral region of
3.2 ppm to 3.3 ppm can also be informative regarding brain tumor charac
teristics. The ratio of phosphocholine to choline is reported to be correlated
with the percentage of highly cellular glioma [301]. Similarly, the ratio of
phosphocholine to choline assessed via in vitro MRS was correlated with the
percentage of malignant tissue in pediatric brain tumors in an investigation by
Tzika et al. [302]. A more recent study [268] among children also corroborates
that significantly higher levels of phosphocholine are found in various types of
cerebellar tumors compared to noninvolved cerebellum. On the other hand,
glycerophosphocholine levels did not differ significantly.
Deeper insights into clinically important molecular dynamics in brain tu
mors are also often provided by in vitro MRS when assessed in direct rela
tion to detailed histopathological analysis. For example, using highresolution
magicangle spinning (HRMAS), Opstad et al. [303] have reported that in as
trocytomas taurine at around 3.3 ppm to 3.4 ppm showed a highly significant
correlation with apoptotic cell density. Moreover, unlike lipids at 0.9 ppm and
1.3 ppm, this association was independent of necrotic tissue. These authors
[303] suggest that taurine may represent an MRvisible biomarker of apoptosis
that is unrelated to the presence of tumor necrosis. Furthermore, they note
that assessment of taurine levels in vivo has been di cult due to overlap with
myoinositol, choline and glucose in that spectral region.
8.2.5 The number of metabolites & non-uniqueness of fitting
As explained earlier, the FFT relies upon postprocessing fitting in attempts
to provide quantification. We elaborated the reasons why this is nonunique.
In short, the number of resonances can only be surmised in this way. Within
cancer diagnostics using in vivo MRS, the consequences of this limitation of
the FFT have been noteworthy. These become particularly troublesome for
closely located and overlapping resonances. With respect to brain metabo
lites, most authors have attributed peaks between 3.8 ppm and 4.0 ppm to
glutamine in the alpha region, and to the second creatine peak. However,
Opstad et al. [145] included glutathione at 2.9 ppm into their LCModel, and
obtained a better fit. They reported that reduced glutathione was significantly
greater in patients with meningiomas compared to those with astrocytomas.