Introduction

Chemical exchange saturation transfer (CEST) MRI is potentially useful for early assessment of treatment response of cancer.¹ However, there are several different acquisition and analysis methods that could impact the quantitative CEST signals. In this study, we investigated e CEST MRI as an imaging biomarker for early treatment response for triple-negative breast cancer (TNBC) patients receiving neoadjuvant chemotherapy (NAC) and compared two saturation power levels and two analysis methods to determine if they affect the CEST signal changes during the patient treatment.

Methods

Our study was approved by the institutional IRB and all patients provided their written informed consent. In total, 13 patients with biopsy-proved TNBC were scanned on a 3T GE human scanner (Discovery MR750) using an 8-channel bilateral breast coil in a prone position. The patients were scanned before the NAC (baseline or BL, N = 13), after 2 cycles of treatment (C2, N = 9), after 4 cycles of treatment (C4, N = 9), and before surgery (presurgery or PS, N = 6). CEST images were acquired using a single-shot fast spin echo (SSFSE) sequence with Field of view Optimized and Constrained Undistorted Single shot (FOCUS). ² The same sequence was scanned first with a CEST saturation power of 0.9 μT and a saturation time of 3500 ms, and then repeated with a CEST saturation power of 2.0 μT and a saturation time of 2000 ms. For both acquisitions, 29 saturation frequencies from -7 to 7 ppm were used in addition to a reference image without CEST saturation. WASSR was acquired in the same scan series for field inhomogeneity correction.³ Fat saturation pulses were employed for fat suppression. Other parameters included FOV = 180 mm × 144 mm, slice thickness = 8 mm and acquisition matrix size = 128 × 128. For both low and high saturation power sequences, TR/TE = 6000/33.6 ms and the total scan time was 4 min 18 sec.

For data acquired at both saturation settings, we measured the CEST signal using MTR_asym averaged between 3.0 – 4.0 ppm (MTR_asym3-4) and Lorentzian line fitting with 2 line shapes that represent water and amide signals at 3.5 ppm. Only the data from 7 to -1 ppm in the Z-spectrum were used in order to minimize the influence of residual fat and NOE signals on the fitting. The magnitude of the fitted amide line was used to represent the CEST signal (Mag3.5). Tumor ROIs were drawn manually on the CEST images by an experienced breast radiologist with reference to the DWI and DCE images. The biopsy marker clip and necrosis were excluded from the measurement ROIs. The changes of the CEST signals for each patient during their treatment were monitored and analyzed.

Results

The MTR_asym maps and their corresponding Z-spectra and MTR_asym are shown for a representative patient from BL to C4 scans in Figs. 1 and 2, respectively. The Mag3.5 maps and fitted lines for the same patient are also shown in Figs. 3 and 4. The average results for all patients are shown in Fig. 5. The Z-spectra were relatively smooth and no obvious fat residual signals or NOE effects were observed. The low power Z-spectra were nosier than the high power spectra (Fig. 2). The high power MTR_asym3-4 was higher than that of the low power at baseline and showed a decreasing trend as the treatment progressed. In comparison, the low power MTR_asym3-4 were similar from BL to C4 (Fig. 2 and 5a). The low power Mag3.5 was higher than that of the high power at baseline and also showed a decreasing trend from BL to C4. However, the high power Mag3.5 had similar values from BL to C4 (Fig. 4 and 5b).

Discussion

Despite being potentially useful for monitoring cancer treatment, there is currently no consensus on the optimal acquisition and analysis methods for CEST imaging. In our study of TNBC during NAC treatment, the MTR_asym in combination with the high saturation power or the Lorentzian line fitting method in combination with the low saturation power demonstrated more consistent CEST signal changes. A potential explanation of our findings may relate to the amide signal strength and lineshape: the amide has a slow exchange rate and hence is expected to produce higher CEST signal with higher saturation power. ⁴ In comparison, lower saturation power produces narrower CEST peaks which are more amenable to line fitting.

Conclusion

Decreasing CEST signals were observed during the NAC treatment of TNBC. However, signal changes were best observed when MTR_asym was used for data acquired at a high saturation power and when Lorentzian line fitting was used for data acquired at a low saturation power.

Acknowledgements

This work was supported in part by the Odyssey Program and Cockrell Foundation Award for Scientific Achievement at The University of Texas MD Anderson Cancer Center (S.Z.).