Breast cancer and non-malignant breast data: Quantification by FPT - Signal Processing in MRS with Biomedical Applications - page 308

appeared to correlate with response to the neoadjuvant therapy.
In a more recent study of 14 women with locally advanced breast cancer
[118], the change in total choline concentrations assessed within 24 hours
after the first dose of chemotherapy was correlated with change in the lesion
size at day 64 (p = 0.001) and differed among those with and without a
response to chemotherapy (p = 0.007). This study was done using a 4T static
magnetic field. All the responders had a fall in total choline concentration
of more than 0.4 mM/kg at 24 hours, whereas the nonresponders all showed
either an increase, no change or at most a fall of 0.2 mM/kg in total choline
concentration at 24 hours.
Tozaki et al. [394] have compared the e cacy of proton MRS with flu
orodeoxyglucose positron emission tomography (FDGPET) for assessing re
sponse of seven patients during the first several days of neoadjuvant chemother
apy for breast cancer. The standardized uptake of FDG showed a significant
correlation with the integral value of the choline peak; this correlation was
particularly strong (r = 0.99, p < 0.001) after chemotherapy. The authors
from Ref. [394] suggest that in vivo MRS may be considered as a potential
alternative for sequential FDGPET evaluation of response to therapy.
10.2.4 Special challenges of in vivo MRS for breast cancer
diagnostics
Notwithstanding the important strides made by in vivo MRS for breast can
cer diagnostics and for assessing response to therapy, there are a number of
challenges that have precluded its more widespread application in this branch
of clinical oncology and in breast radiology.
Proton MRS for breast cancer diagnostics has generally required suppres
sion of lipid, since the MR spectra from the breast are usually dominated by
fat. This hinders localized shimming and also produces sideband artefacts
[392]. Lipid suppression has most often been achieved by increasing the echo
time. However, this leads to diminished signal intensity. Hu et al. [391] have
suggested the possibility of suppressing lipid via an echofilter using short echo
times. However, Stanwell and Mountford [392] emphasize that suppression of
the lipid resonance eliminates the possibility of evaluating lipid that is part
of the actual disease process.
Overall, use of various echo times impedes consistent interpretation of data
from in vivo MRS of the breast [392], and this may be clinically important.
Metabolites with short T 2 relaxation times will have decayed at longer TE,
e.g., myoinositol whose estimated concentrations best distinguished breast
cancer from fibroadenoma in our analysis [21, 22, 25] of in vitro MRS data
[395] (see the next subsection).
Another major problem with the current applications of in vivo MRS for
breast cancer diagnostics is the reliance upon the composite choline peak. This
compromises diagnostic accuracy, since choline may also be observed in benign
breast lesions, as noted in a number of the cited results. On the other hand, it