Abstract
BACKGROUND: An accurate assessment of isocitrate dehydrogenase (IDH) status in patients with glioma is crucial for treatment planning and is a key factor in predicting patient outcomes. In this study, we investigated the potential value of whole-tumor histogram metrics derived from synthetic magnetic resonance imaging (MRI) in distinguishing IDH mutation status between astrocytoma and glioblastoma. METHODS: In this prospective study, 80 glioma patients were enrolled from September 2019 to June 2022. All patients underwent pre- and post-contrast synthetic MRI scan protocol. Immunohistochemistry (IHC) staining or gene sequencing were used to assess IDH mutation status in tumor tissue samples. Whole-tumor histogram metrics, including T1, T2, proton density (PD), etc., were extracted from the quantitative maps, while radiological features were assessed by synthetic contrast-weighted maps. Basic clinical features of the patients were also evaluated. Differences in clinical, radiological, and histogram metrics between IDH-mutant astrocytoma and IDH-wildtype glioblastoma were analyzed using univariate analyses. Variables with statistical significance in univariate analysis were included in multivariate logistic regression analysis to develop the combined model. Receiver operating characteristic (ROC) and area under the curve (AUC) were used to assess the diagnostic performance of metrics and models. RESULTS: The histopathologic analysis revealed that of the 80 cases, 41 were classified as IDH-mutant astrocytoma and 39 as IDH-wildtype glioblastoma. Compared to IDH-wildtype glioblastoma, IDH-mutant astrocytoma showed significantly lower T1 [10th percentile (10th), mean, and median] and post-contrast PD (10th, 90th percentile, mean, median, and maximum) values as well as higher post-contrast T1 (cT1) (10th, mean, median, and minimum) values (all P<0.05). The combined model (T1-10th + cT1-10th + age) was developed by integrating the independent influencing factors of IDH-mutant astrocytoma using the multivariate logistic regression. The diagnostic performance of this model [AUC =0.872 (0.778-0.936), sensitivity =75.61%, and specificity =89.74%] was superior to the clinicoradiological model, which was constructed using age and enhancement degree (AUC =0.822 (0.870-0.898), P=0.035). CONCLUSIONS: The combined model constructed using histogram metrics derived from synthetic MRI could be a valuable preoperative tool to distinguish IDH mutation status between astrocytoma and glioblastoma, and subsequently, could assist in the decision-making process of pretreatment.