0690

Assessment of Intracranial Metastases in Patients using Hyperpolarized ¹³C MRI

Casey Y Lee^1,2, Hany Soliman³, Benjamin J Geraghty^1,2, Nadia D Bragagnolo^1,2, Albert P Chen⁴, William J Perks⁵, Arjun Sahgal³, Michael W Chan⁶, Sean Symons⁷, and Charles H Cunningham^1,2
¹Medical Biophysics, University of Toronto, Toronto, ON, Canada, ²Physical Sciences, Sunnybrook Research Institute, Toronto, ON, Canada, ³Radiation Oncology, Sunnybrook Health Sciences Centre, Toronto, ON, Canada, ⁴GE Healthcare Technologies, Toronto, ON, Canada, ⁵Pharmacy, Sunnybrook Health Sciences Centre, Toronto, ON, Canada, ⁶Diagostic Imaging, Trillium Health Partners, Mississauga, ON, Canada, ⁷Radiology, Sunnybrook Health Sciences Centre, Toronto, ON, Canada

Synopsis

Hyperpolarized ¹³C MRI was used to acquire images of [1-¹³C]lactate and ¹³C-bicarbonate from the injected [1-¹³C]pyruvate in 8 patients with brain metastases. Lesions were manually contoured and the mean tumor ¹³C-lactate signal was converted to a z-score by extending the approach previously described in Lee et al. (2019). As expected, the z-score ranks of the anatomical regions were less concordant in patients compared to controls. A range of lactate z-scores were observed in metastatic lesions, showing metabolic heterogeneity consistent with the known heterogeneity in metastatic features and clinical status. The lesions with the highest and 5^th highest lactate z-scores progressed.

Introduction

Brain metastases develop in upwards of 40% of cancer patients¹. Radiation therapy is a standard treatment approach, requiring MRI to monitor changes in lesion size in order to assess treatment response. Significant clinical heterogeneity in brain metastases patients makes it difficult to predict treatment response and changes in lesion size can take several months to manifest². Hyperpolarized (HP) ¹³C MRI³ is a new imaging approach that has the potential to predict radiation treatment response at an earlier time point by probing the changes in tumour metabolism. In particular, tumour lactate concentration is known to be positively correlated with metastatic potential and radiation resistance^4,5.

The consistent ¹³C-lactate signal topography of the human brain following the injection of HP [1-¹³C]pyruvate has been previously demonstrated^6,7. The strong background ¹³C-lactate signals from brain parenchyma confounds the measurement of ¹³C-lactate signal from metastatic lesions. In this study, we extended the segmentation-based approach previously⁶ applied to characterize ¹³C-lactate signals from control human brains to normalize the lactate signal from metastatic lesions by converting the ¹³C-lactate signals into z-scores. The tumour lactate z-scores were compared to the primary tumour type, lesion location and the treatment response status.

Methods

Image acquisition Written informed consent was obtained from patients with one or more newly developed brain metastases (n=8). Prior to being positioned head-first in a GE MR750 3.0T MRI scanner (GE Healthcare, Waukesha, WI), a 20-gauge intravenous catheter was inserted into the patient’s forearm. The patient’s head was secured in the support for head-coil base for the standard 8-channel neurovascular receive array (Invivo Inc.). This support can be docked with either the 8-channel ¹H array or a home-built single-tuned ¹³C birdcage without moving the patient’s head.

At the start of the exam, ¹H localizer data and ¹H multi-echo reference data were acquired using the scanner’s body coil. The ¹³C birdcage was put in place and the dose of HP [1-¹³C]pyruvate was injected (0.1mmol/kg dose at 4mL/s followed by a 25mL normal saline flush at 5mL/s). Lactate, bicarbonate and pyruvate images were sequentially acquired using spectral-spatial excitation⁸ and 3D dual-echo EPI readout⁹ (5s temporal resolution; axial prescription; 1.5cm-isotropic resolution with FOV 24x24x36cm³). For anatomical reference, T1-weighted images were acquired using fast spoiled GRE (axial, FOV 25.6x25.6cm², 1mm-isotropic resolution, TR 7.6ms, TE 2.9ms, flip angle 11°) with the ¹H array.

Image Reconstruction ¹³C images were reconstructed offline using MATLAB R2013a (The MathWorks Inc., Natick, MA) using a data-driven approach to correct for geometric distortion⁹. The final ¹³C images were generated by summing the odd and even echoes and summing the individual timepoints over the full 60s acquisition window to provide maximal SNR.

Structural Segmentation Structural segmentation of the ¹³C images was performed using the LONI pipeline processing environment^6,10. The original ¹³C images were resampled to match the coordinates and matrix size of the T1-weighted images in MATLAB and segmented into the 56 structural regions contained in the LPBA40 atlas using the LONI pipeline. For each subject, the mean ¹³C signal from each atlas region was converted to a z-score by subtracting the mean of all regions and dividing by the standard deviation across regions for that subject.

Each metastatic lesion ROI was manually contoured on the T1-weighted images by a radiation oncologist. The mean ¹³C-lactate signal for each lesion was then converted to a z-score using the mean and standard deviation of the 56 atlas regions.

Results and Discussion

Selected examples of ¹³C-lactate, ¹³C-pyruvate, and ¹³C-bicarbonate images are shown in Fig. 1. For each patient, lactate z-scores from each LPBA40 atlas region and the lesion ROIs are illustrated in Fig. 2 and corresponding information is summarized in a table in Fig. 3. As expected, the ranking of z-scores showed weaker concordance for this patient group compared to the previously reported values from 14 control participants⁶ (W = 0.66 and 0.83, respectively). A range of lactate z-scores (-1.29 to 3.76) was observed amongst different lesions, as well as between lesions in a single patient.

The lesion (IMET0012a) with the highest z-score (3.76) progressed after stereotactic radiosurgery. Lesion IMET0034b had a lactate z-score = 0.93 (5^th highest) progressed and was surgically resected at 1 month post-imaging. All of the lesions with lactate z-score below the median were either stable or responsive. The relationship between lactate z-score, lesion location, current clinical status, and primary tumour type are shown in Fig. 4.

In future work, a total of 10 patients with brain metastases will be scanned under the current protocol and these patients’ clinical progression will be monitored up to 6 months post ¹³C-imaging to evaluate the potential role of lactate z-score in characterizing and predicting treatment outcomes for individual lesions.

Conclusion

A segmentation-based approach was used to calculate lactate z-scores for metastatic lesions in patients with brain metastases. A range of lactate z-scores were observed in metastatic lesions, showing metabolic heterogeneity consistent with the known heterogeneity in metastatic features and clinical status. The lesions with the highest and 5^th highest lactate z-score progressed, and all of the lesions with lactate z-score below the median were either stable or responsive.

Acknowledgements

The authors thank Julie Green and Sumeet Sachdeva for coordinating the study, and Ruby Endre and Garry Detzler for MR technical support. Funding support from the Brain Canada and Canadian Institutes for Health Research.

References

1. DeVita, Vincent T., Theodore S. Lawrence, and Steven A. Rosenberg, eds. DeVita, Hellman, and Rosenberg's cancer: principles & practice of oncology. Vol.

2. Lippincott Williams & Wilkins, 2008.Lin, Xuling, and Lisa M. DeAngelis. "Treatment of brain metastases." Journal of clinical oncology 33.30 (2015): 3475.

3. Ardenkjær-Larsen, Jan H., et al. "Increase in signal-to-noise ratio of> 10,000 times in liquid-state NMR." Proceedings of the National Academy of Sciences 100.18 (2003): 10158-10163.

4. Walenta, Stefan, et al. "High lactate levels predict likelihood of metastases, tumor recurrence, and restricted patient survival in human cervical cancers." Cancer research 60.4 (2000): 916-921.

5. Quennet, Verena, et al. "Tumor lactate content predicts for response to fractionated irradiation of human squamous cell carcinomas in nude mice." Radiotherapy and Oncology 81.2 (2006): 130-135.

6. Lee, Casey Y., et al. "Lactate topography of the human brain using hyperpolarized 13C-MRI." NeuroImage 204 (2020): 116202.

7. Grist, James T., et al. "Quantifying normal human brain metabolism using hyperpolarized [1–13C] pyruvate and magnetic resonance imaging." NeuroImage 189 (2019): 171-179.

8. Cunningham, Charles H., et al. "Pulse sequence for dynamic volumetric imaging of hyperpolarized metabolic products." Journal of magnetic resonance 193.1 (2008): 139-146.

9. Geraghty, Benjamin J., et al. "Dual‐Echo EPI sequence for integrated distortion correction in 3D time‐resolved hyperpolarized 13C MRI." Magnetic resonance in medicine 79.2 (2018): 643-653.

10. Rex, David E., Jeffrey Q. Ma, and Arthur W. Toga. "The LONI pipeline processing environment." Neuroimage 19.3 (2003): 1033-1048.

Figures

Fig. 1) Selected example images of ¹³C-lactate,¹³C-bicarbonate, and ¹³C-pyruvate are shown for lesions with ‘high’, ‘intermediate’, and ‘low’ lactate z-scores and the case which metastatic lesion surrounded a cyst. IMET#### indicates patient ID followed by an alphabet indicating ROI ID (see Fig. 3).

Fig. 2) The distribution of lactate z-scores for each patient (n=8) is plotted against the LPBA40 atlas region labels. Kendall’s coefficient of concordance (W) amongst patient group was 0.66. Lactate z-scores for each metastatic lesion are also plotted. Lesion features are summarized in Fig. 3.

Fig. 3) Summary of patient, type of primary tumour and lactate z-score, location and status of each contoured region. *Abbreviations: p = progressed; s = stable; r = responding; t = previously treated lesion; na = treatment response status ‘not yet available’. (^&For IMET0034b, metastatic lesion was found at the periphery of a cyst.)

Fig. 4) Lactate z-score from metastatic lesions are grouped based on lesion location (left), clinical status (middle), and primary tumour types (right). Individual markers correspond to the same type of markers used in Fig. 2 to distinguish different lesions.

Proc. Intl. Soc. Mag. Reson. Med. 28 (2020)

0690