Imaging scoring systems for preoperative molecular diagnoses of lower-grade gliomas

Recent advance in molecular characterization of gliomas showed that patient prognosis and/or tumor chemosensitivity correlate with certain molecular signatures; however, this information is available only after tumor resection. If molecular information is available by routine radiological examinations, surgical strategy as well as overall treatment strategy could be designed preoperatively.With the aim to establish an imaging scoring system for preoperative diagnosis of molecular status in lower-grade gliomas (WHO grade 2 or 3, LrGGs), we investigated 8 imaging features available on routine CT and MRI in 45 LGGs (discovery cohort) and compared them with the status of 1p/19q codeletion, IDH mutations, and MGMT promoter methylation. The scoring systems were established based on the imaging features significantly associated with each molecular signature, and were tested in the another 52 LrGGs (validation cohort).For prediction of 1p/19q codeletion, the scoring system is composed of calcification, indistinct tumor border on T1, paramagnetic susceptibility effect on T1, and cystic component on FLAIR. For prediction of MGMT promoter methylation, the scoring system is composed of indistinct tumor border, surface localization (FLAIR), and cystic component. The scoring system for prediction of IDH status was not established. The 1p/19q score ≥ 3 showed PPV of 96.2% and specificity of 98.1%, and the MGMT methylation score ≥ 2 showed PPV of 77.4% and specificity of 67.6% in the entire cohort.These scoring systems based on widely available imaging information may help to preoperatively design personalized treatment in patients with LrGG.