3911
Strategy to overcome the spectral overlap between the 2-hydroxyglutarate and lipid resonances at 2.25 ppm
Pegah Askari1, Ivan E Dimitrov1, Michael Levy1, Toral R Patel1, Edward Pan1, Bruce E Mickey1, Craig R Malloy1, Elizabeth A Maher1, and Changho Choi1
1University of Texas Southwestern Medical Center, Dallas, TX, United States
The lipid resonance at 2.25 ppm often complicates 1H MRS evaluation of 2HG. Incorporating new lipid basis sets in spectral fitting of 3T PRESS data from 43 glioma patients showed complete distinction between IDH mutation and wildtype (accuracy, sensitivity, and specificity all unity).
FIG 4. (A,B) ROC analyses of the 2HG estimates from 43 patients are presented for Fitting methods 1 and 2. A red circle on an ROC curve corresponds to the smallest distance to the upper-left corner of the curve, at which a cutoff value was obtained as 1.3 and 0.8 mM for Fitting methods 1 and 2, respectively. (C,D) 2HG estimates from 24 IDH mutated and 19 IDH wildtype tumor patients are shown for Fitting methods 1 and 2. Green lines indicate the cutoff values from the ROC curves. For Fitting method 2, the 2HG estimates in IDH mutated tumors were 0.8 - 6.8 mM and those in IDH wildtype tumors were 0 - 0.4 mM.
FIG 1. Representative in vivo PRESS TE 97 ms spectra from three glioma patients are presented with LCModel fitting outputs and voxel positioning on T2-FLAIR images. LCModel-returned 2HG signals are shown with 2HG estimates and CRLB in brackets. The LCModel built-in lipid basis set (Fitting method-1) included Lip09, Lip20, and Lip13. New lipid basis set (Fitting method-2) included LipNew1, LipNew2, and Lip13. Dotted lines denote exclusion of the lipid signals in the basis set. Spectra are scaled with respect to the water signal from the voxel. Vertical lines are drawn at 2.25 and 0.9 ppm.