0397
T1, T2 relaxometry for tissue cell density quantification in glioma imaging: Exploration study via 11C-methionine PET and its validation via stereotactic tissue sampling1Neurosurgery, Osaka International Cancer Institute, Osaka, Japan, 2Canon Medical Systems Corporation, Tochigi, Japan, 3Diagnostic and Interventional Radiology, Osaka International Cancer Institute, Osaka, Japan, 4Institute for Clinical Research, Osaka National Hospital, Osaka, Japan, 5Nuclear Medicine and Tracer Kinetics, Osaka University Graduate School of Medicine, Suita, Japan, 6Diagnostic Pathology and Cytology, Osaka International Cancer Institute, Osaka, Japan, 7Neurosurgery, Osaka University Graduate School of Medicine, Suita, Japan
Synopsis
In this study, the authors attempted to use T1, T2 relaxometry for predicting tumor cell density within the brain in glioma patients. The study was conducted in two stages, first as an exploratory study comparing T1, T2 relaxometry and 11C-methinonie PET images, and second as a validation study using intraoperative stereo-tactically obtained tissues. The authors were able to identify a range of T1 and T2 relaxation time indicative of high cell density, which finding was confirmed by stereotactic tissue sampling. This technique was further able to create predictive tumor cell density map by T1, T2 relaxometry alone.
PURPOSE
Visualization of tumor load into the brain tissue is a crucial process in glioma treatment. It has been suggested, however, that conventional T1-, T2-weighted and contrast enhanced MRI is insufficient to obtain this information and metabolic imaging, such as amino acid PET, is desired to complete pre-treatment imaging study. It should also be noted, however, that these kinds of imaging are not easily accessible world-wide, and a more practical technique should be developed to (semi-)quantitively visualize tissue tumor load in glioma patients. In this study, the authors attempted to achieve this objective using T1, T2 relaxometry mapping in glioma patients. The study was conducted in two stages, first as an exploratory study comparing T1, T2 relaxometry and 11C-methinonie PET images, and second as a validation study using intraoperative stereo-tactically obtained tissues (Figure 1). The developed technique was also challenged to visualize tissue cell density by T1, T2 relaxometry.METHODS
Patient cohort (Study 1): Eight patients harboring gliomas (Low grade glioma 5, High grade glioma 3) were included. T1, T2 relaxometry and 11C-methionine PET were available in all 8 cases.
Patient cohort (Study 2): 12 patients harboring high grade glioma were included. 34 tissues were stereo-tactically obtained under image guidance using a BrainLab neuronavigational system. T1 and T2 relaxometry maps were also available.
T1 and T2 relaxometry: MP2RAGE and multi-echo T2-weithted images were acquired using either a 3T (MAGNETOM Prisma, Siemens) or 1.5T (MAGNETOM Aera, Siemens) (Figure 1). These images were converted into T1 and T2 relaxometry using Olea Sphere, Olea Nova+ (Canon Medical Systems). A correction factor between 1.5 and 3T data was obtained using a normal volunteer brain scan data.
Lesion segmentation for Study 1: Lesion segmentation for analysis was handcrafted in 3D including the entire T2WI abnormal area. Defined lesion was then segmented into either area with high methionine uptake (tumor/normal ratio >1.5, assigned as Met-PET high) or low uptake (≤ 1.5, assigned as Met-PET low), which cut-off is considered reasonable in previous studies (1).
Tissue cell density measurement and image/tissue co-registration: The number of cells was counted on Hematoxylin Eosinstaining. The locations of tissue sampling were identified on T1 and T2 relaxometry maps using the Dicom coordinates store in the navigation system which was created at the time of stereo-tactic tissue sampling. No manual adjustment was performed ensuring objectiveness of the study.
Tumor load imaging via T1, T2 relaxometry: Estimation of high probable cell density tissue was calculated from the data obtained in study 1, combining both data obtained by T1 and T2 relaxometry.
RESULTS
Exploration study (Study 1): As in Figure 2, Met-PET high and low segments were separable in both T1 and T2 relaxometry. T1 relaxation time between 2000 and 3200 msec (in 3T) was indicative of Met-PET high, which can be interpreted as high cell density area. Similarly, T2 relaxation time between 125 and 225 msec (in 3T) was indicative of Met-PET high area. Met-PET high probability was highest at 1500 msec in T1 and 180 msec in T2 relaxometry (Figure 2).
Validation study (Study 2): Tissue obtained from above mentioned predictive high cell density area in both T1 and T2 relaxometry showed statistically significantly higher tissue cell density than tissues obtained from predictive low cell density area (Figure 3).
Cell density imaging via combined T1 and T2 relaxometry: Predictive cell density imaging based on T1 and T2 relaxometry showed reasonable correlation with 11C-methionine PET as can be seen in Figure 4. It should be noted that two types of T2 hyper-intense lesions was recognized; i.e. that with high methionine uptake along with predicted high cell density via T1 and T2 relaxometry and the other with low methionine uptake along with predicted low cell density. Furthermore, when signals deriving from tissues with T1 relaxation time shorter than 2000 msec was suppressed via inversion recovery technique, the latter type of T2 hyper-intense lesion was eliminated leaving the former being visually recognized.
DISCUSSION
This investigation tested the hypothesis that T1 and T2 relaxometry can be used as image surrogate to predict tissue tumor load in glioma imaging. This hypothesis was founded on the assumption that T1 and T2 relaxation time inhabit information on the tissue micro architecture(s) in glioma. In fact, previous report did suggest that T2 relaxation time between 125 and 250 msec can be suggested as areas of non-enhancing tumor lesions (2). The current study was able to confirm this observation and was able to further expand this concept to using T1 relaxation time as supplementary information with histological confirmation via stereotactic tissue sampling.Acknowledgements
This research was supported by JSPS, Kakenhi, MSD life science foundation, Takeda life science foundation and the Uehara Memorial FoundationReferences
(1) Kinoshita M, Arita H, Okita Y, et al. Comparison of diffusion tensor imaging and 11C-methionine positron emission tomography for reliable prediction of tumor cell density in gliomas. J Neurosurg. 2016;125(5):1136-1142.
(2) Ellingson BM, Lai A, Nguyen HN, Nghiemphu PL, Pope WB, Cloughesy TF. Quantification of Nonenhancing Tumor Burden in Gliomas Using Effective T2 Maps Derived from Dual-Echo Turbo Spin-Echo MRI. Clin Cancer Res. 2015;21(19):4373-4383.
Figures
