0308

Brain Tumors Disrupt the Resting-State Connectome
Darian Hadjiabadi1, Leland Pung1, Jiangyang Zhang2, BD Ward3, Woo-Taek Lim1, Meghana Kalavar2, Nitish V Thakor1, Bharat B Biswal4, and Arvind P Pathak1,2,5

1Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD, United States, 2Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, MD, United States, 3Department of Biophysics, The Medical College of Wisconsin, Milwaukee, WI, United States, 4Department of Biomedical Engineering, New Jersey Institute of Technology, Newark, NJ, United States, 5Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD, United States

Synopsis

Resting-state functional MRI (rsfMRI) has become indispensable for mapping the changes in ‘connectivity’ between brain regions in a range of diseases including brain tumors. However, the complex interplay between abnormal brain tumor vasculature, tumor blood flow, and cancer cell-induced neurovascular uncoupling can confound the interpretation of resting-state connectivity in patients. Therefore, in this preclinical study we quantified brain tumor-induced changes on resting-state connectivity relative to that in healthy brains, followed by histological validation. RsfMRI revealed that brain tumors alter the resting-state connectome, and histology confirmed that this was largely due to cancer cell-induced disruption of the neurovascular unit.

Purpose

Resting state functional MRI (rsfMRI) has become indispensable for mapping the changes in ‘connectivity’ of spatially distinct brain regions1 in a wide array of disease models including brain tumors2. However, the complex interplay between the abnormal brain tumor vasculature, anomalous tumor blood flow3, and cancer cell-induced neurovascular uncoupling4, can confound the acquisition and interpretation of resting-state connectivity in patients2. Since little is known about how the blood-oxygenation-level dependent (BOLD) rsfMRI signal is modulated by the presence of a brain tumor, in this preclinical study we quantified brain tumor-induced changes on the rsfMRI signal and resting state connectivity, relative to that in healthy brains. Histological validation revealed that brain tumor-induced disruption of the neurovascular unit, as well as brain tumor progression were responsible for the observed in vivo changes in resting state connectivity.

Methods

9L-GFP brain tumor cells were orthotopically inoculated into the cortices of SCID mice (n=10) as described in5. Brains of healthy SCID mice (n=10) served as the control group. Animals were imaged in vivo on a 400 MHz Bruker spectrometer using the following sequences after localized shimming: (i) T2w rapid acquisition with relaxation enhancement (RARE), RARE-factor=8, TE = 15.0ms, TR = 3.5s, NA=8, resolution = 0.1mm×0.1mm, 16-24 coronal slices, slice = 0.3mm; (ii) gradient-echo EPI, TE = 8.4ms, TR =  400ms, 110 repetitions, resolution = 0.2mm×0.2mm, 16-24 coronal slices, slice= 0.3mm. Animals were imaged under isoflurane anesthesia with body temperature maintained at 37°C and respiration rate at 40-60bpm. After in vivo MRI, animals were perfused with an intravascular tracer (dextran-TRITC, 70kDa), euthanized, brains excised, fixed, frozen and cut for immunofluorescent labeling of neurovascular unit components, such as the astrocyte marker–glial fibrillary acidic protein (GFAP). A 3D mouse atlas6 was used as a reference for segmenting anatomical regions-of-interest (ROI) using Amira® (FEI Software, OR). ROI include left/right: hippocampus(Hi), neocortex(Neo), olfactory bulb(OB), thalamus(Th), striatum(Str), hypothalamus(Hy), and brainstem(Stem). AFNI software was utilized for all image processing7. EPI data were co-registered to the anatomical data, followed by Gaussian 0.5 mm FWHM spatial filtering and 0.01 Hz high-pass filtering. Resting-state connectivity between ROI pairs was evaluated as the cross-correlation (CC) coefficient between their spatially-averaged BOLD signals. Resting-state maps were obtained through cross-correlation analysis of the average resting-state BOLD time course for a given ROI with all the voxels for each comparison ROI. For group-wise comparisons, CC’s for ROI pairs were averaged across each group. Additionally, the resting-state ‘connectome’ of healthy and tumor-bearing mouse brains were represented by force-directed spatial graphs using the Kamada-Kawai algorithm8. Finally, the power spectrum was computed from the average BOLD time-course for each ROI in the tumor and healthy brain.

Results

Resting-state maps show that brain tumors significantly alter the resting-state connectivity in the right hippocampus (Fig.1b, 1d). Negative correlations observed in the left hippocampus of the healthy brain (Fig.1b) were attenuated by the presence of a brain tumor (Fig.1d). For healthy brains, ipsilateral ROI exhibited positive correlations, whereas contralateral ROI exhibited negative correlations (Fig.2a). In contrast, connectivity between brain ROI was reorganized for tumor-bearing brains (Fig.2b). Overlaid Kamada-Kawaii plots (Fig.2e) highlight the altered connectome of tumor-bearing brains (Fig.2d) relative to healthy brains (Fig.2c). The global impact of the brain tumor on connectivity was also evident via attenuation of the resting-state BOLD signal between ROI occupying right (Fig.3a) and left hemispheres (Fig.3b). Finally, histology revealed sparse astrocytic-vascular coverage within the brain tumor (Fig.4b) relative to that of healthy brain regions (Fig.4a), thereby validating the presence of neurovascular uncoupling within tumor-associated regions.

Discussion

Although rsfMRI has been widely used for presurgical mapping and other applications in brain tumor patients, factors such as cancer cell-induced neurovascular uncoupling can confound the utility of this approach. Recent preclinical rsfMRI tumor studies have primarily focused on BOLD fluctuations arising from abnormal tumor hemodynamics9,10. To the best of our knowledge, this is the first study to characterize the global effect of brain tumor progression on the rsfMRI signal. We discovered that brain tumor-induced alterations in resting-state connectivity were attributable to neurovascular uncoupling as validated by histology. Surprisingly, we observed that progressing brain tumors induced a global reorganization of the resting-state network that extended to the hemisphere contralateral to the tumor, accompanied by global diminution of BOLD signal fluctuations. This work highlights the local and global effects of the brain tumor on its microenvironment, and suggests potential approaches for making rsfMRI a useful cancer biomarker. For example, one could envision ‘normalization’ of the brain tumor vasculature with antiangiogenic therapy, followed by restoration of the ‘brain tumor connectome’ to the ‘heathy brain connectome’ as detected by rsfMRI.

Acknowledgements

Supported by NCI 1R01CA196701-01 and 1R21CA175784-02.

References

1. Biswal, B. et. al (1995). Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magnetic Resonance in Medicine, 34(4), 537-541.

2. Agarwal, S. et. al (2016). Neurovascular uncoupling in resting state fMRI demonstrated in patients with primary brain gliomas. Journal of Magnetic Resonance Imaging ,43, 620–626.

3. Jain, R. K. et. al (2007). Angiogensis in brain tumors. Nature Neuroscience, 8, 610-622.

4. Lee, Jisook et. al (2009). Glioma-induced remodeling of the neurovascular unit. Brain Research, 1288, 125-134.

5. Pathak, A.P. et. al (2001). MR-derived cerebral blood volume maps: Issues regarding histological validation and assessment of tumor angiogenesis. Magnetic Resonance in Medicine, 46(4), 735-747.

6. Aggarwal, M. et. al (2009). Magnetic resonance imaging and micro-computed tomography combined atlas of developing and adult mouse brains for stereotaxic surgery. Neuroscience, 162(4), 1339-1350.

7. Cox, R. W. (1996). AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers and Biomedical Research, 29(3), 162-173.

8. Kamada, Tomihisa and Kawai, Satoru. (1989). An algorithm for drawing general un-directed graphs. Information Processing Letters, 31, 7-15.

9. Baudelet, C. et. al (2006). The role of vessel maturation and vessel functionality in spontaneous fluctuations of T2*-weighted GRE within tumors. NMR in Biomedicine, 19(1), 69-76.

10. Gonclaves, Miguel R. et. al (2015). Decomposition of spontanteous fluctuations in tumor oxygenation usng BOLD MRI and independent component analysis. British Journal of Cancer, 113(8), 1168-1177.

Figures

Figure 1: Visualization of alterations in resting-state connectivity induced by a medium sized brain tumor. (a) Healthy brain label field overlaid on an anatomical slice, showing left hippocampus (orange arrow) and right hippocampus (white arrow). (b) Right hippocampus resting-state map overlaid on an anatomic slice for a normal brain. (c) Tumor-bearing label field overlaid on an anatomical slice showing tumor (red with white “T”), left hippocampus (orange arrow), and right hippocampus (white arrow). (d) Right hippocampus resting-state map overlaid on an anatomical slice for a tumor-bearing brain. Tumor location is indicated by white hatched outline.

Figure 2: Brain tumors disrupt inter- and intra-hemispheric resting-state connectivity. Correlation coefficient (CC) matrices illustrating the resting state functional connectivity for: (a) ROIs from healthy mice; (b) ROIs from brain tumor-bearing mice. Kamada-Kawai (KK) plots corresponding to the CC matrices in (a) and (b). (c) Average KK plot for normal mouse brains, (d) average KK plot for tumor-bearing mouse brains, and (e) an illustration of the alterations in connectivity between tumor-bearing and normal brains by overlaying the KK-plots in (c) and (d). Labels have been omitted in (e) for clarity.

Figure 3: Tumors suppress BOLD fluctuations across brain regions. Blood Oxygenation Level Dependent (BOLD) periodograms dissect total BOLD signal variance into respective frequency contributions. Periodograms of (a) right hippocampus and (b) left hippocampus were acquired from respective averaged BOLD time course of healthy (gray) and tumor-bearing (red) subjects. Frequency range is 0.01-0.078 Hz.

Figure 4: The neurovascular unit is disrupted in brain tumors. Optical microscopy images of tissue sections from healthy and brain tumor-bearing murine brain. (a) Hippocampal region from a healthy mouse brain in which astrocytes have been labeled with GFAP (green channel) and perfused blood vessels with dextran-TRITC (red channel). (b) Tumor (T) and surrounding region in which astrocytes have been labeled with GFAP (green channel) and blood vessels with dextran-TRITC (red channel). The absence of GFAP labeling within the tumor region demonstrates disruption of the neurovascular unit in brain tumor-bearing brains. Scale bar = 50 µm in both panels.

Proc. Intl. Soc. Mag. Reson. Med. 25 (2017)
0308