Stereotactic radiosurgery (SRS) is common for the management of brain metastases. It is a localized treatment which has led to increased patient survival [1] and reduced side-effects, but volume changes may not manifest for months. Chemical exchange saturation transfer (CEST) shows great promise for improving the characterization of brain tumours and their response to therapy [2-10]. Our objective was to assess the utility of CEST metrics in characterizing the tumour response to SRS. To limit a priori assumptions about which feature of CEST might be sensitive to therapy, a CEST spectrum was collected at equally spaced frequency offsets. This study explores not only whether CEST changes correlate with the change in tumour volume, but whether CEST imaging at early time-points is predictive of future growth or regression of the tumour.

The CEST spectrum was approximated by five pools: "magnetization transfer" (broad lineshape at 0 ppm), "bulk water" (0 ppm), "amide" (3.5ppm), "amine" (2ppm), and "NOE" (-3.5ppm). Two groups of CEST metrics were considered for each CEST peak:

1) The magnetization transfer ratio (MTR) evaluated at the frequency of the pool:

$$MTR\left(\Delta\right)=\frac{S_{ref} -S(\Delta)}{S_{ref}}$$

2) The peak amplitude of each CEST effect (with amplitude “A”, width “w”, and offset frequency “Δ₀”) from a Lorentzian decomposition of the CEST spectrum:

$$S(\Delta) = 1 - \sum_{i=1}^5A_i\left(1 + \left(\frac{\Delta-\Delta_{0i}}{0.5*w_i}\right)^2\right)^{-1}$$

20 patients with brain metastases were scanned on a 3T Philips Achieva at three time-points: pre-treatment, one week and one month post-therapy. CEST imaging was optimized to be sensitive to small changes in the CEST pool. Image acquisition: 2D, single-shot gradient echo EPI (FE-EPI), TR: 1532ms, effTE = 30ms, resolution: 2.5mm, slice thickness: 3mm. Saturation scheme: 750ms block pulse, 0.52µT, preceding EPI. Offset frequencies: -750 to 750 Hz with reference at 100,000 Hz. Averages: 5, Total Scan Time: 12 minutes. Post-Gd T1-weighted images were used as a reference to define ROIs in enhancing tumour, and ipsilateral normal-appearing white matter (iNAWM). The mean MTR was calculated over the ROI at three different offset frequencies: the amide, amine and NOE peak offsets. CEST peakfit metrics were determined for each voxel, then averaged over the ROI with a partial-volume correction [12].

MTR metrics offered the best discrimination between tumour and adjacent white matter at the pre-treatment time-point (Fig. 1). Except for the amide and amine peak amplitude, metrics were all higher in the iNAWM.

Predictive factors were identified by comparing CEST changes between pre-treatment and one week with volume changes between pre-treatment and one month. Several significant trends were observed (Fig. 2), including a positive correlation between volume and Amide MTR (R=0.50, p=0.085) in tumour. This corroborates research that shows cell lines which are responsive to radiation have a reduced APT effect [2] after treatment, and that the APT of apoptotic/necrotic tissue is reduced [13]. Decrease in NOE peak amplitude in the iNAWM at one week was also correlated (R=0.73, p=0.0048) with volume reduction at one month. One theory is that if NOE can be considered correlated with cell proliferation rate, then this reflects radiosensitivity of the iNAWM, and possibly by extension, the tumour[14]. No significant correlation was observed between the change in T1 after one week and the volume changes at one month.

Prognostic factors were also identified by comparing the pre-treatment CEST metrics to the volume changes at one month, including MT amplitude in tumour (R=-0.5, p=0.094), and NOE amplitude (R=-0.64, p=0.0180), amide width (R=-0.57, p=0.042) and NOE MTR (R=-0.51, p=0.080) in iNAWM (Fig. 3). Interestingly, the correlation was greater in the iNAWM than within the tumour. This is likely a consequence of the iNAWM ROIs being larger and more homogeneous, forming a much more consistent baseline upon which to assess altered metabolism [15]. Conclusion: Early CEST metric increases in either tumour or the surrounding iNAWM following SRS correlated with future tumour expansion. At the pre-treatment stage, low MT amplitude in tumour, or low NOE amplitude or MTR in iNAWM correlated with worse outcome at one month, which could be investigated as a biomarker for poor response to SRS or vulnerability to future metastases.

Early CEST metric increases in either tumour or the surrounding iNAWM following SRS correlated with future tumour expansion. At the pre-treatment stage, low MT amplitude in tumour, or low NOE amplitude or MTR in iNAWM correlated with worse outcome at one month, which could be investigated as a biomarker for poor response to SRS or vulnerability to future metastases.