Hypoxia is assumed to play an important role in the prognosis and therapy response of glioblastoma.¹ Therefore, a robust method for imaging of hypoxia is urgently wanted. In recent years, MRI methods based on the blood oxygenation level dependent (BOLD) effect have been thoroughly investigated ^2,3 and a multi-parametric method for the measurement of an apparent relative oxygen extraction fraction (rOEF) has been successfully applied in glioma patients.⁴ However, a number of confounding influences like susceptibility artifacts, orientation effects within white matter and systematic errors due to model inadequacies impede straightforward application.⁵ Since direct validation with hypoxia-related PET tracers like 18F-FMISO ⁶ is difficult,⁷ this study aims to shed some light on the potential significance of BOLD based oxygenation measures in glioma, by exploring the characteristics of multi-parametric rOEF measurements in a sample of 36 mostly high grade glioma patients from a multimodal MR/PET study.

45 patients (57.8±16.8, 26 men) with suspected glioma (30 °IV (GBM), 5 °III, 7 °II, 3 other) underwent a simultaneous MRI and dynamic ¹⁸F-FET-PET examination on a clinical 3 T Biograph mMR scanner (Siemens Medical Solutions). The advanced clinical MRI protocol comprised R2' mapping (voxel size 2x2x3 mm³, matrix 128x128, 30 slices) by separate acquisition of a multi-gradient echo (12 echoes, TE1 = 5 ms, TE = 5 ms, TR = 1950 ms, α = 30°, rapid flyback, acq. time 4:08min) and a multi-echo TSE sequence (8 echoes, TE₁ = 16 ms, TE = 16 ms, TR = 4040 ms, acq. time 5:04 min). Relative cerebral blood volume (rCBV) was obtained by dynamic susceptibility contrast (DSC) imaging (single-shot GE EPI: TR = 1500 ms, TE = 30 ms, α = 90°, 60-80 dynamics) during a bolus injection of 15 ml Gd-DTPA (prebolus of 7.5 ml).⁸ Data processing used SPM8 (www.fil.ion.ucl.ac.uk/spm) and custom programs in Matlab (MathWorks). T2* evaluation included correction for motion and magnetic background gradients and T2 fitting was restricted to even echoes.⁵ $$$rOEF=\frac{R2'}{c\cdot rCBV}$$$ was calculated from $$$R2'=\frac{1}{T2^*}-\frac{1}{T2}$$$ and rCBV using $$$c=4/3\cdot π\cdot γ\cdot ∆χ\cdot B_0 = $$$ 317 Hz at 3T.⁵ Volumes of interest (VOI) were defined with Vinci (http://www.nf.mpg.de/vinci3) as follows: Control - unaffected contralateral tissue; Edema - very bright FLAIR signal; T2T - FLAIR-visible solid appearing tumor; CET - tumor tissue with T1 contrast enhancement; rOEF > μ+σ - tumor areas with high rOEF values (larger than sum of mean (µ) and standard deviation (σ) of rOEF in the control region). Special care was taken to exclude artifacts (areas with necrosis, bleeding, iron deposition, macroscopic susceptibility perturbation, Fig. 1).

rOEF maps with diagnostic quality could be obtained in 36 patients (26 GBM, 5 °III, 5 °II). Fig.1 shows images from a GBM patient with a large necrosis and edema, illustrating VOI selection and artifact exclusion. Fig. 2 depicts the relative volumes of the evaluated VOIs (2a) together with the patient averages of rOEF (2b), rCBV (2c) and R2' (2d) obtained within these VOIs for glioma of different grade. Compared to Edema and solid appearing tumor with FLAIR signal alterations (T2T), areas with contrast enhancement (CET) and supra-threshold rOEF values (rOEF > μ+σ) are relatively small. This is also due to the fact necrotic and hemorrhagic areas were excluded. While the lowest rOEF values occur within CET, within each VOI, rOEF shows a clear trend to be higher in GBM than °III and °II glioma (Fig. 2b). Comparison with rCBV (Fig. 2c) and R2' (Fig. 2d) behavior suggests that the difference within VOIs is mainly influenced by R2' while between VOIs rCBV appears to be the dominating factor. While rCBV appears to be distinctly different between VOIs, especially between Edema and CET, but similar across tumor grades, R2' clearly varies with tumor grade especially within edema.

Despite the painstaking exclusion of any area with artificial, i.e. non-oxygenation related, R2' enhancement, these results clearly confirm previous findings in human glioma where rOEF values were found to increase with tumor grade within similar VOIs.⁵ However, unlike in this previous study,⁵ we did not find a clear increase of the supposedly hypoxic tumor area, i.e. the area with supra-threshold rOEF values, with tumor grade. In our opinion, this can mostly be explained by the exclusion of any central necrosis which represents a large hypoxic area in most high grade glioma. According to these results, high rOEF values, supposedly corresponding to a high oxygen extraction, prevail in edematous areas with low rCBV. Whether this also translates into tissue hypoxia, clearly needs further investigation.