For a better understanding of the antiangiogenic Bevacizumab (BVZ) therapy and to identify patients who might benefit from it, we used ¹H MRS to monitor metabolic changes during treatment. Up to now most of the research employing MRS has been focused on the effects of BVZ on the tumor’s energy and membrane metabolites as potential markers ^1–3. A different metabolite that can be measured with ¹H-MRS is myoinositol (MI). MI plays an important role in the cellular osmoregulation of the brain and is predominantly produced by astrocytes ⁴. Its intracellular concentration changes depending on the extracellular osmolality ^5–7. BVZ has a strong impact on tumor vessels, the blood-brain barrier and thus most likely on the osmotic environment ⁸. We therefore hypothesized that a change of the MI concentration should be measurable during BVZ therapy. To investigate our hypothesis we evaluated the spectroscopic data of 30 prospectively evaluated patients with recurrent glioblastoma before and during therapy with BVZ. Another 9 patients with recurrent glioblastoma receiving CCNU (Lomustine) and VM26 (Teniposid) chemotherapy served as a control group.Patients with a histological diagnosis of glioblastoma and radiologically confirmed recurrence were included. All participants underwent an MRS examination before the start of therapy and at regular MRI follow up. Follow ups were performed after 8 weeks for BVZ patients and 10-12 weeks for CCNU/VM26 patients (Patient characteristics shown in Table 1). The detailed MRS protocol can be found in the previously mentioned publication by Hattingen et al. ². The data was sampled from voxels within the tumor ROI and the control ROI in the contralateral hemisphere. LCModel was employed to analyze the MI signal intensity. To evaluate the potential of MI as a predictive marker for BVZ therapy we performed an overall survival (OS) analysis for the BVZ cohort.MI concentrations in the tumor were lower compared to control tissue (p< .001) for both cohorts. Concentrations increased significantly during BVZ therapy in tumor (p<.001) and control tissue (p=.001), but not during CCNU/VM26 treatment (Figure 1). For the BVZ cohort, higher MI concentrations in the control tissue at baseline (p=.021) and higher differences between control and tumor tissue (delta MI, p=.011) were associated with longer survival. A Kaplan-Meier analysis showed a median OS of 164 days for patients with a deltaMI < 1,817 mmol/l and 275 days for patients with a deltaMI > 1,817 mmol/l (Figure 2). No differences were observed for the relative changes or the post treatment concentrations (data shown in Table 2).

A low MI concentration in the tumor compared to the control tissue is in accordance with previous MRS studies ^9–11. The VEGF secretion of glioblastoma causes a disruption of the blood-brain barrier followed by a leakage of electrolytes and small molecules from the vessels, affecting the osmolality of the extracellular environment ¹². The distinct reduction of contrast enhancement and peritumoral edema during BVZ therapy suggests a normalization of the osmotic environment, which is reflected by the increase of the MI concentration in the tumor. In contrast, we did not observe changes of the MI concentrations during CCNU/VM26 therapy, indicating that the cytotoxic treatment does not affect the regulation of intracellular MI.

While tumor MI concentrations were not significant, we identified baseline control MI and delta MI levels to be predictive for overall survival. As in the tumor, control MI was increased after BVZ treatment. This either suggests an impact of BVZ on healthy tissue or, alternatively, a tumor related decrease of MI at baseline with a therapy effect comparable to the one in the tumor. Since low baseline values in the control tissue and a lower difference to the values in the tumor were negative predictors for overall survival, they might indicate a more widespread effect of the tumor. Consequently this would indicate an affection of the contralateral control tissue. We assume that this observation could either be explained by direct tumor cell Infiltration, which has already been described to occur in more than 50% of untreated tumors by Matsukado et al. in 1961 ¹³ or an indirect tumor effect. Concerning indirect effects, a spreading edema that is not yet visible in MRI scans or increased intracranial pressure affecting the contralateral hemisphere have to be taken into account.

Pre-therapeutic MI concentrations are predictive of overall survival in patients with recurrent glioblastoma treated with BVZ. Our data confirm that recurrent glioblastoma show a strong metabolic reaction to BVZ. Further, our results support the hypothesis that MI might be a marker for early tumor cell invasion.