Comparison of amide proton transfer metrics for IDH genotype discrimination in gliomas at low B1.

Amide proton transfer-weighted (APTw) imaging enables tumor characterization by detecting mobile proteins and peptides. Low B1 levels provide greater spectral selectivity, advantageous for resolving APT signals. However, low B1 also increases the T1 dependence of conventional APT metrics, including magnetization transfer ratio asymmetry (MTRasym), multipool Lorentzian fitting, and inverse difference, despite their different strategies for mitigating confounding effects. This study compared the newly proposed quasi-steady-state (QUASS)-based apparent exchange-dependent relaxation (AREX) with conventional APT metrics for discrimination of IDH genotypes in gliomas under finite radiofrequency irradiation with B1 of 0.75 μT, a level previously shown to provide good tumor-to-normal tissue contrast. Protein phantoms with varying concentrations were constructed, and glioma patients were prospectively enrolled and classified according to IDH mutation status. QUASS was used to reconstruct steady-state Z-spectra, from which APT metrics were quantified. Statistical analyses included the Mann-Whitney U test, area under the curve (AUC), and DeLong test, with false discovery rate correction. In the phantom study, QUASS-based AREX showed the strongest correlation with egg white concentration among the evaluated APT metrics (r = 0.988, P < 0.001). Among 45 patients (22 IDH-mutant and 23 IDH-wildtype), QUASS-based AREX significantly differentiated IDH genotypes (P = 0.007), whereas MTRasym, multipool Lorentzian fitting, and inverse difference did not (P = 0.733, 0.250, and 0.087, respectively). QUASS-based AREX achieved the highest AUC of 0.771, significantly outperforming the other metrics (all P < 0.05). Under the saturation parameters used in this study, QUASS-based AREX provided better IDH genotype discrimination than the evaluated conventional APT metrics.