As anticipated, our results confirmed that quantitative relaxometry using SyMRI could differentiate the IDH-mutant gliomas leading to an increased sensitivity compared to the qualitative T2-FLAIR mismatch sign. Qualitative results showed high specificity; however, sensitivity was low and inter-rater agreement was fair. Quantitative results showed that there were significant differences in the 10–90th percentiles and the mean between the gliomas for all SyMRI parameters. The ROC analysis revealed that the T2-value of the 10th and 50th percentiles, and the mean showed the highest diagnostic ability in differentiating gliomas. In the simulated T2-FLAIR mismatch sign, T2-FLAIR mismatch present cases showed a higher mismatch ratio compared to that of the absent cases.
Previous studies have reported that the T2-FLAIR mismatch sign has been shown to predict astrocytoma, IDH-mutant with 95–100% specificity [14, 15]; however, sensitivity has been low at 46–51% [8, 14, 16]. Inter-rater agreement has been shown to have a wide range of κ = 0.38–0.88 [5, 8, 14, 15]. Our results are consistent with these previous results. Low sensitivity is probably due to the binary scoring system used, whereas subtle changes may result in different interpretations across readers [8]. Some readers in this study also reported the T2-FLAIR mismatch in oligodendroglioma; In this case, a cystic change was present in the tumor. Since T2-FLAIR mismatch occurs based on the long T1-relaxation time, the intratumoral cystic component simulates a T2-FLAIR mismatch. The T2-FLAIR mismatch false positive results have been reported in previous studies [15, 17]. Therefore, surgical specimens are still needed to diagnose gliomas instead of relying on the T2-FLAIR mismatch sign alone.
Our quantitative results showed long T1- and T2-relaxation times and an increased PD within the astrocytoma compared to the oligodendroglioma. To our knowledge, this is the first study to quantitatively evaluate IDH-mutant gliomas using SyMRI. These prolonged T1- and T2-relaxation times imply the result of fluid changes in the IDH-mutant astrocytoma. Deguchi et al. revealed that abundant microcysts were observed upon hematoxylin-eosin staining of specimens from T2-FLAIR mismatched region in the astrocytoma with IDH-mutant [15], which may reflect T1- and T2-relaxation time prolongations. Kinoshita et al. evaluated gliomas quantitatively using MP2RAGE images calculated via Bayesian inference modeling and showed that T2-FLAIR mismatch region exhibited extremely long T1- and T2-relaxation times [16]. Their results also support fluid changes within the IDH-mutant astrocytoma. Our quantitative results showed that there were significant differences in the mean value between the gliomas for all SyMRI parameters. These findings may support clinical utility of SyMRI to diagnose gliomas because we require a simple ROI to place in the tumor, not requiring a histogram analysis. Further, as expected, quantitative evaluation significantly increased the sensitivity compared to the qualitative T2-FLAIR mismatch sign. Advanced MR techniques such as dynamic perfusion MRI [18] and magnetic resonance spectroscopy [19] have also revealed differences between the IDH-mutant and wildtype gliomas. However, dynamic perfusion MRI requires contrast media and 2-hydroxyglutarate magnetic resonance spectroscopy have been shown to be associated with false positive cases with intratumoral hemorrhage [20]. In addition, 2-hydroxyglutarate detection with magnetic resonance spectroscopy is a technically challenging due to the spectral overlap of 2-hydroxyglutarate with background metabolites [21].
Our result of the simulated T2-FLAIR mismatch sign (% change) suggested that radiologists were able to detect the difference between T2WI and FLAIR only when it was approximately > 50%. A previous study reported that astrocytoma, IDH-mutant did not always reveal the T2-FLAIR mismatch sign (only 22%) [5]. Kinoshita et al. tested that shortening the TI could contribute to better visualization of the T2-FLAIR mismatch signs and suggested that fine-tuning inversion recovery sequence may improve the detection in IDH-mutant gliomas [16]. Our quantitative evaluation using SyMRI could overcome this low-sensitivity and inter-observer variability. Contrary to expectations, the tumor size showed the opposite result owing to the sample size in the simulated study, thereby, resulting in a large variation.
This study has several limitations. First, the sample size was small, and our study included postoperative cases. Although postoperative changes may affect the results of the qualitative evaluation, the quantitative evaluation was still useful to differentiate the gliomas in this study. Second, our study did not evaluate the IDH-wildtype astrocytomas. It would be desirable to proceed with a subsequent study including patients with astrocytoma, IDH-wildtype. Third, we did not include the whole tumor volume in the histogram analysis. Instead, we used the maximum section of the tumor, with its boundary defined by the hyperintensity on T2WI. However, in previous studies on the T2-FLAIR mismatch, the whole-volume histogram analysis has not been performed and only the maximum section of the tumor has been utilized.