Introduction

Chemo-radiation treatment for glioblastoma¹ affects normal brain tissues and leads to significant decline in patients’ quality of life (QoL) and neuro-cognitive function. Treatment-induced changes in normal brain are not detectable on clinically used anatomical imaging early after the treatment^2,3. However their impact on QoL appear several months after end of the treatment³. Magnetization transfer (MT) is sensitive to treatment-induced changes in white matter integrity and microstructure and might detect such damages early. Identification of the patients who are sensitive to treatment and are expected to experience deteriorated QoL may help in minimizing these effects by adjusting the treatment or additional medication.

Methods

Subjects: 14 newly diagnosed GBM patients were recruited (REB approved) and MRI-scanned immediately before start of the treatment (Day₀). Patients were treated with 30 fractions of 2Gy radiation (delivered via IMRT) + daily Temozolomide (TMZ) over 6 weeks. 12 patients (out of 14) were also MRI-scanned one month after end of the treatment (Day₇₀).

MRI: was performed on a 3T Philips Achieva scanner. An axial slice passing through the tumour was imaged with TFE (TR/TE=1971/29ms, TFE-factor=2, Matrix=144×144, FOV=20×20cm, slice=3mm).
qMT: Imaging covered offsets between 2 and 780ppm with 14 offsets (with equal logarithmic scale spacing). RF saturation consisted of 2 & 4 block pulses of 242.5ms each with 2.5ms gaps interleaved (485 & 970ms total) and B1=1.5,3,5 µT^4,5.
The six MT spectra were fitted to the two-pool MT model proposed by Henkelman⁶ in transient state, and the four model parameters [RM_0B/R_A,T_2B,R,1/(R_AT_2A)] were calculated where RM_0B is magnetization transfer rate; M_0B, the macromolecular pool-size; R, the exchange rate from MT to water pool; R_A, the longitudinal relaxation rate of the water pool; and T_2A and T_2B, the transverse relaxation times of the water and MT pools, respectively.
Analysis: qMT/T₁/B₁/postGd-T_1w/FLAIR images were co-registered to the first acquired MT image. Normal appearing white matter (NAWM) ROI was defined manually on postGd-T_1w and FLAIR, avoiding tumour mass, edema and areas of white matter hyperintensity. The total radiation dose (over 6-week treatment) in radiotherapy plan was divided into three dose levels (XRT<25Gy, 25Gy<XRT<45Gy, XRT>45Gy). Metrics were calculated voxelwise and median value over ROI was used.

Results & Discussions

Day₀ scan of one patient was discarded due to imaging problems, thus eleven patients had both Day₀ and Day₇₀ scans. Figure 1a-b show the NAWM ROI of an illustrative patient overlaid on postGd-T_1w and T_2w-FLAIR. Radiotherapy plan (XRT) used for treatment is shown in Fig.1c, showing the three dose level regions. Figure 1d shows the overlap of NAWM region and radiotherapy plan which was used in calculating summary statistics.
Figure 2 presents MT parametric maps for the patient at Day₀ and Day₇₀ showing RM_0B is decreasing and Fig.3 shows the histogram distribution of two qMT parameters averaged over all patients at each scan.
Table 1 reports qMT parameters segregated for the three dose levels. For XRT<25Gy region there is statistically significant reduction in RM_0B (which represents the magnetization transfer rate) after treatment. RM_0B reduction shows loss of white matter integrity that may result in cognitive dysfunction. When moving closer to the tumour at Day₀ (25Gy<XRT<45Gy and XRT>45Gy) value of RM_0B is already lower than XRT<25Gy (which is the farthest region from the tumour). This might be due to the fact that GBM cells invade their surrounding normal tissues and destroy them (such destruction is not visible on clinical MRI).
Figure 4 shows RM_0B for each of 11 patients that had both Day₀ and Day₇₀ scans. For XRT<25Gy region approximately half the patients (6/11) experienced significant reduction in RM_0B while the other half experience no change. Thus, treatment impacts patients differently where some patients are resistant to treatment and others experience significant damage to NAWM (Similar trends with smaller magnitudes are occurring for 25Gy<XRT<45Gy region).
For XRT>45Gy region that received the highest radiation dose, MT parameter was similar before and after treatment (RM_0B,Day0=2.22±0.17 vs. RM_0B,Day70=2.21±0.19) indicating NAWM was already damaged by the tumour infiltration and did not change with treatment.
Similarly for 25Gy<XRT<45Gy, RM_0B at Day₀ had a value between the other two dose levels which indicates there was some NAWM damage. After treatment (Day₇₀) however, RM_0B of 25Gy<XRT<45Gy region decreased to values similar to the other dose levels (RM_0B,Day70= 2.25±0.10, 2.23±0.17, 2.21±0.19 for the three dose levels).
The parameter 1/(R_AT_2A) represents direct water effect and the fact that it did not change between Day₀ and Day₇₀ shows the above mentioned changes were due to change in MT properties and not just tissue T₁ and T₂. This also re-confirms the fact that these subtle early changes in NAWM cannot be detected on clinical MRI scans that rely on T₁ and T₂.

Conclusions

Chemo-radiation damages NAWM and its impact on patients’ quality of life becomes apparent several months to years after treatment. MT is sensitive to microstructural changes in white matter and might detect such changes early. As early as one month post-treatment, we observed MT changes at NAWM that were far from tumour and received very low radiation dose. More importantly, not all patients experienced the same damage to NAWM after treatment. Early diagnosis of the patients that are sensitive to treatment might guide further medication to preserve quality of life.

Acknowledgements

This study was funded by Canadian Institute of Health Research (CIHR PJT156252) and Terry Fox Research Institute (TFRI project 1034).