Characterization of the inflammatory post-ischemic tissue by full volumetric analysis of a multimodal imaging dataset.

Introduction: In vivo positron emission tomography (PET) and magnetic resonance imaging (MRI) support non-invasive assessment of the spatiotemporal expression of proteins of interest and functional/structural changes. Our work promotes the use of a volumetric analysis on multimodal imaging datasets to assess the spatio-temporal dynamics and interaction of two imaging biomarkers, with a special focus on two neuroinflammation-related biomarkers, the translocator protein (TSPO) and matrix metalloproteinases (MMPs), in the acute and chronic post-ischemic phase.

Aim: To improve our understating of the neuroinflammatory reaction and tissue heterogeneity during the post ischemic phase, we aimed (i) to assess the spatio-temporal distribution of two radiotracers, [F]DPA-714 (TSPO) and [F]BR-351 (MMPs), (ii) to investigate their spatial interaction, including exclusive and overlapping areas, and (iii) their relationship with the Tw-MRI ischemic lesion in a transient middle cerebral artery occlusion (tMCAo) mouse model using an atlas-based volumetric analysis.

Methods: As described by Zinnhardt et al. (2015), a total of N = 30 C57BL/6 mice underwent [F]DPA-714 and [F]BR-351 PET-CT and subsequent MR imaging 24-48 h (n = 8), 7 ± 1 days (n = 8), 14 ± 1 days (n = 7), and 21 ± 1 days (n = 7) after 30 min transient middle cerebral artery occlusion (tMCAo). To further investigate the spatio-temporal distribution of [F]DPA-714 and [F]BR-351, an atlas-based ipsilesional volume of interest (VOI) was applied to co-registered PET-CT images and thresholded by the mean uptake + 2.5*standard deviation of a contralateral striatal control VOI. Mean lesion-to-contralateral ratios (L/C), volume extension (V in voxel), percentages of overlap and exclusive tracer uptake areas were determined. Both tracer volumes were also compared to the lesion extent depicted by Tw-MR imaging.

Results: Both imaging biomarkers showed a constant small percentage of overlap across all time points (14.0 ± 14.2%). [F]DPA-714 reached its maximum extent and uptake at day 14 post ischemia (V = 12,143 ± 6262 voxels, L/C = 2.32 ± 0.48). The majority of [F]DPA-714 volume (82.4 ± 16.1%) was exclusive for [F]DPA-714 and showed limited overlap with [F]BR-351 and Tw-MRI lesion volumes. On the other hand, [F]BR-351 reached its maximum extent already 24-48 h after tMCAo (V = 7279 ± 4518 voxels) and significantly decreased at day 14 (V = 1706 ± 1202 voxels). Focal spots of residual activity were still observed at day 21 post ischemia (L/C = 2.10 ± 0.37). The majority of [F]BR-351 volume was exclusive for [F]BR-351 (81.50 ± 25.07%) at 24-48 h and showed 64.84 ± 28.29% of overlap with [F]DPA-714 from day 14 post ischemia while only 9.28 ± 13.45% of the [F]BR-351 volume were overlapping the Tw-MRI lesion. The percentage of exclusive area of [F]DPA-714 and [F]BR-351 uptakes regarding Tw-MR lesion increased over time, suggesting that TSPO and MMPs are mostly localized in the peri‑infarct region at latter time points.

Conclusion: This study promotes the use of an unbiased volumetric analyses of multi-modal imaging data sets to improve the characterization of pathological tissue heterogeneity. This approach improves our understanding of (i) the dynamics of disease-related multi-modal imaging biomarkers, (ii) their spatiotemporal interactions and (iii) the post-ischemic tissue heterogeneity. Our results indicate acute MMPs activation after tMCAo preceding TSPO-dependent (micro-)gliosis. The spatial distribution of MMPs and gliosis is regionally independent with only minor (< 20%) overlapping areas in peri‑infarct regions.