Objective:

To explore the clinical value of features of radiomics model based on magnetic resonance imaging (MRI) in preoperative predicting the prognosis of patients with glioblastoma.

Methods:

A total of 130 glioblastoma patients who underwent preoperative MRI were enrolled in the study. Radiomics features were extracted from the edema area

the core area and the necrotic core area

of the tumor respectively. Least absolute shrinkage and selection operator (LASSO) regression algorithm was implemented to establish the radiomics characteristic model

which was tested by ten-fold cross validation method. Patients were divided into high and low-risk groups according to the optimum threshold of radiomics score. Subgroups analysis was performed to observe whether temozolomide adjuvant radiotherapy can prolong the survival of high-risk patients.

Results:

A total of 16 features were selected to construct the radiomic model significantly related to overall survival for patients with glioblastoma

including 4 from the edema area

9 from the core area

3 from the necrotic core of glioblastoma. Multivariate Cox regression analysis showed that radiomics score was an independent risk factor for patients with glioblastoma. According to the threshold of radiomics score

patients was categorized as low risk (n=69

53.1%) and high risk (n=61

46.9%). There were significant differences in survival and progression free survival between the two groups. Subgroup analysis showed that high-risk patients the OS (14.2 months) with temozolomide concurrent chemoradiotherapy was significantly longer than that in the radiotherapy group (7.1 months

P

=0.000 16).

Conclusion:

The radiomics features obtained from MRI-based radiomic model are associated with the prognostic and can stratify patients into high-risk and low-risk groups successfully. High risk patients can benefit from temozolomide concurrentradiotherapy.