|
Predicting
progression in children with pediatric brain tumors using magnetic resonance
neuroimaging
A A.
Tzika, Loukas G. Astrakas, David Zurakowski, Maria K. Zarifi, Tina P. Young,
Liliana Goumnerova, Douglas C. Anthony, Mark Kieran, Nancy J. Tarbell, Peter M.
Black
Massachusetts
General Hospital, Boston, MA; Children's Hospital Boston, Boston, MA;
Dana-Farber Cancer Institute, Boston, MA; Department of Pediatric Oncology,
Boston, MA
Introduction. In searching for novel neuroimaging methods, we have become interested in the
ability of proton magnetic resonance (MR) spectroscopic imaging to provide an
independent source of information, in addition to MR imaging, about tumor
metabolism and response to treatment.
MR perfusion may also be useful in assessing brain tumor response to
treatment.
In this study, we sought to test the hypothesis that changes in choline ratios
obtained by proton MR spectroscopic imaging (MRSI) combined with perfusion MRI
can differentiate between clinically stable and progressive outcomes in
pediatric brain tumors.
Patients
and Methods. Fifty MRSI neuroimaging exams were performed on 27 children with
neuroepithelial brain tumors during therapy using a 1.5-T MR system.
Each exam was rated as stable or progressive by clinical and imaging
criteria.
Data sets of 1 cc. to 1.2 cc were obtained with proton MRSI using a multi-voxel
method.
Perfusion MRI was performed using an MRI sequence during a compact bolus of
paramagnetic contrast agent (Gd-DTPA).
A dose of Gd-DTPA at 0.1mmol/kg (0.2cc/kg) was administered intravenously using
an automatic injector with a power injector rate of 1-2 cc/sec.
MR spectroscopic derived parameters including choline-containing compounds (Cho)
and n-acetylasparate (NAA) were measured.
Parameters were normalized by dividing values from within tumor regions to the
mean value of total creatine (tCr) in normal tissue.
Relative cerebral blood volume (rCBV) values were obtained from perfusion MRI
and normalized to the middle cerebral artery (MCA) territory of the same
patient.
The odds ratio (OR) and 95% confidence intervals (CI) were determined by
logistic regression analysis.
Results. Significant differences were found between progressing and stable outcomes
in percent change of normalized Cho (P<0.05).
Percent change in Cho/NAA was significantly higher in the progressive (n=18)
compared to stable (n=32) outcomes (P<0.001).
Logistic regression indicated that % change in Cho/NAA was the most accurate
predictor of tumor progression (likelihood ratio test = 33.4, P<0.001).
Using a 20% increase as a cutoff, Cho/NAA correctly classified 16/18 progressing
cases (sensitivity = 0.89, 95% CI: 0.65-0.99) and 28/32 stable cases
(specificity = 0.88, 95% CI: 0.71-0.97).
The risk of tumor progression was over 50 times higher in cases with a 20% or
greater increase in Cho/NAA (OR=55, lower bound 95% CI: 9.2).
Cho and rCBV were positively correlated (r=0.61, P<0.01), and rCBV was higher
in progressing tumors (P=0.03).
Conclusion. Indices determined by proton MR spectroscopic imaging, particularly Cho/NAA,
offer prognostic value in evaluating neuroepithelial brain tumors.
Together with Cho/NAA, perfusion MRI may provide additional information
regarding tumor progression.
Copyright
© 2003 American Association for Cancer Research. All rights reserved.
Source: http://aacr03.agora.com/planner/displayabstract.asp?presentationid=10068
|
