Pitfalls in 2HG Detection With TE-Optimized MRS at 3T

In vivo MRS of 2-hydroxyglutarate (2HG) may provide diagnostic and monitoring biomarkers in isocitrate dehydrogenase (IDH)-mutated glioma. A previous meta-analysis has shown good diagnostic accuracy of TE-optimized PRESS for IDH-mutated glioma, but most studies feature IDH-wildtype glioma as a comparison. However, when considering newly identified brain lesions that may mimic glioma, full characterization of its diagnostic utility should also consider the accuracy of 2HG measurement in nontumor tissue. Therefore, we tested how well TE-optimized 2HG levels distinguish between IDH-mutated glioma and nontumor tissue-here, normal-appearing brain. We further examined the impact of different spectral modeling strategies (baseline stiffness, macromolecule inclusion, and basis set composition).

Forty-eight patients with diagnosed/suspected IDH-mutated glioma were enrolled. 3T MRS data were acquired from tumor and contralateral nontumor tissue with TE-optimized PRESS and analyzed with "LCModel" software. ROC analysis evaluated 2HG estimates' ability to distinguish IDH-mutated glioma from nontumor tissue. Modeling interactions between 2HG and other metabolites were evaluated to identify reasons for potential false-positive 2HG detection.

TE-optimized PRESS distinguished IDH-mutated glioma from nontumor tissue with lower sensitivity (range 0.76-0.62) and specificity (0.85-0.78) than the literature suggests for IDH-mutated versus IDH-wildtype glioma. Strong negative correlations between gamma-aminobutyric acid (GABA) and 2HG persisted across all modeling strategies and may lead to false-positive 2HG detection in nontumor tissue. We further present a cautionary example from a patient on a ketogenic diet, showing that ketone body acetone can interfere with 2HG detection.

Spectral overlap with GABA and acetone can lead to false-positive 2HG detection in nontumor tissue.