| Sumario: | Background: H3K27-altered diffuse midline gliomas (H3K27-altered DMGs) are classified as WHO grade 4 tumors despite their histopathological characteristics. However, histopathological grades are known to have an important effect on prognosis. This study aims to investigate the prognostic impact of histopathological grades on H3K27-altered DMGs and to predict the grades using multiparametric MRI. Methods: We conducted a retrospective study on H3K27-altered DMG patients diagnosed between December 2016 and February 2023. Tumors were classified as histologically lower- or high-grade DMGs by experienced pathologists. The influence of histopathological grades on prognosis was analyzed by Kaplan-Meier curve and Cox regression. Propensity score matching minimized baseline bias. Clinical and multimodal MRI variables were analyzed through age-stratified subgroup analysis. Logistic regression model was constructed for grade prediction and evaluated using the C-index and F1 score. Results: 90 (53 lower-grade, 37 high-grade) had survival data and 44 (26 lower-grade, 18 high-grade) had complete multiparametric MRI data. Lower-grade patients exhibited significantly longer median overall survival than high-grade (10.5 months vs. 4 months, p < 0.05). Histopathological grade was confirmed as an independent prognostic factor. Qualitative MR imaging features revealed that high-grade DMGs in juveniles showed significantly larger enhancement quality and edema proportion (p < 0.05). Quantitative perfusion parameters like relative cerebral blood flow (rCBF) and relative cerebral blood volume (rCBV) differed significantly between high- and lower-grade groups in both juveniles and adults (p < 0.05). rCBF_99th, enhancement quality, and edema proportion, selected from variables with intergroup differences, formed a nomogram model with a C-index of 0.8596, AUC of 0.885 (0.833 validation), and F1 score of 0.8648. Conclusions: Histological grades could not be ignored when assessing the prognosis of H3K27-altered DMGs. Multimodal MR imaging signatures may play an important role in the prediction of histological grades and thus contribute to a layered risk management of DMGs.
|
|---|