Loukas G. Astrakas PhD, Abstract Co-Author: Nothing to Disclose

David Zurakowski PhD, Abstract Co-Author: Nothing to Disclose

Maria Zarifi, Abstract Co-Author: Nothing to Disclose

Karen Marcus MD, Abstract Co-Author: Nothing to Disclose

Nancy Jane Tarbell MD, Abstract Co-Author: Nothing to Disclose

A. Aria Tzika PhD, Presenter: Nothing to Disclose

The objective of this study was to test the hypothesis that combining information from biologically important intracellular molecules, obtained by proton MR spectroscopic imaging (MRSI) and histology, will increase the probability of predicting survival in children with pediatric CNS tumors.

Proton MRSI was performed using multi-voxel chemical shift imaging with point resolved spectroscopy (PRESS) and volume preselection on 76 children with neuroepithelial brain tumors prior to therapy on a 1.5-T MR system. The following acquisition parameters were used: TR=1s, TE=65msec, 16x16 phase encoding matrix, 160 mm FOV, slice thickness of 10 mm, 1250 Hz spectral width, 2 averages and 512 points. Data sets of 1-1.2 cc nominal resolution were obtained. Patients who died (n=18) and those alive (n=58) were compared with respect to median (interquartile range) levels of choline and lipids/lactate using the Mann-Whitney U-test. Association between WHO grade and survival was determined by chi-square. Area under the ROC curve was calculated as a measure of diagnostic performance for choline and lipid/lactate in differentiating patient survival. Multiple stepwise logistic regression analysis was applied to identify independent predictors of survival considering choline, lipid/lactate, and WHO grade.

ROC curve analysis indicated that discrimination of survival outcome for both choline (AUC = 0.725, 95% confidence interval = 0.600 – 0.859) and lipid/lactate (AUC = 0.687, 95% confidence interval 0.544 – 0.830). Multiple stepwise logistic regression analysis indicated that among choline, lipid/lactate, and WHO grade, only choline was an independent predictor of survival (likelihood ratio test = 9.64, P = 0.002). Lipid/lactate (P = 0.16) and WHO grade (P = 0.58) provided no additional information in differentiating alive vs. dead outcome beyond that provided by choline.

Our data suggest that proton MRSI biomarkers for predicting survival is an excellent test in children, and may aid in the diagnosis and help facilitate development of therapies tailored to the biological behavior of the tumor.

Astrakas, L, Zurakowski, D, Zarifi, M, Marcus, K, Tarbell, N, Tzika, A, Prediction of Survival in Children with Brain Tumors Based on Proton Magnetic Resonance Spectroscopic Imaging.  Radiological Society of North America 2005 Scientific Assembly and Annual Meeting, November 27 - December 2, 2005 ,Chicago IL. http://archive.rsna.org/2005/4411116.html