Magnetic resonance elastography for the quantitative assessment of post-treatment glioblastoma using a multiparametric imaging protocol

In glioblastoma, the most aggressive primary brain tumor in adults, treatment effect is commonly assessed with contrast- enhanced magnetic resonance imaging (MRI), but is challenged by pseudoprogression and pseudoresponse. Pseudoprogression is a transient post-radiotherapy increase in the contrastenhancing lesion, which mimics tumor progression, whereas pseudoresponse describes a reduction in contrast enhancement after antiangiogenic treatment, without a true decrease in tumor size. 

Quantitative imaging biomarkers non-invasively measure normal and pathological tissue features, potentially improving cancer management through detection, tumor characterization and treatment response assessment. Before clinical adoption, Quantitative imaging biomarkers require technical and biological validation to establish measurement consistency and to investigate their potential link with tumor biology or treatment-induced changes.

Magnetic resonance elastography (MRE) is an advanced MRI technique used to quantify tissue viscoelastic properties such as the complex shear modulus magnitude |G*| (here referred to as stiffness) and its components the storage modulus G’ and loss modulus G’’. The repeatability of MRE has been established in healthy volunteers, but not yet in patients. Although MRE has shown potential for characterizing glioblastoma, treatment-related changes in tumor viscoelastic properties have so far only been investigated in preclinical studies.

This thesis consists of 3 sub-studies, which evaluated MRE repeatability and interobserver agreement in glioblastoma (sub-study 1), the ability of MRE to detect changes in tumor viscoelastic properties after antiangiogenic treatment(sub-study 2) and radiotherapy (sub-study 3).

Methods
The repeatability of MRE was assessed using test-retest data, obtained in one preoperative imaging session and independently assessed by 2 observers. To investigate the ability of MRE to detect post-treatment changes in tumor and peritumoral viscoelasticity, pre-treatment measurements were compared with measurements acquired at 2 post-treatment time points. The temporal evolution of viscoelastic properties was assessed at the cohort level and according to time from treatment start-stop due to progression (sub-study 2) and progression free survival (PFS) after radiotherapy (sub-study 3).

Results
In sub-study 1, MRE reliably detected tumor stiffness and storage modulus changes larger than 0.10 kilopascals (kPa) or 8.11%, with excellent interobserver agreement. Larger measurement variability was observed in tumor loss modulus, normal-appearing white matter (NAWM) and normalized tumor/NAWM values.

No overall cohort-level changes could be detected after antiangiogenic treatment in sub-study 2. Patients with long time to progression had significantly higher stiffness and storage modulus than those progressing early, particularly at the first post-treatment scan. Differences in stiffness and storage modulus according to O6-methylguanine-DNA methyl-transferase (MGMT) promotermet hylation status were also observed. After radiotherapy (sub-study 3), longer PFS was associated with lower tumor and peritumoral |G*| and G’ values. Patients with PFS>6 months showed post-treatment decreases in tumor stiffness and storage modulus and lower values compared to patients with PFS<6 months.

Conclusion
The results of this thesis suggest that MRE provides repeatable and interobserver consistent measurements of glioblastoma viscoelastic properties, particularly for stiffness and storage modulus. Post-treatment changes were mostly detectable in stiffness and storage modulus after stratification by progression timing, with the largest differences observed at the first posttreatment scan. Viscoelastic property changes appeared to differ between treatment types, suggesting that MRE may capture treatment-induced changes.