Histologically Validated Ex Vivo Characterization of Intraplaque Hemorrhage and Lipid Components Using Quantitative Susceptibility Mapping

BACKGROUND AND PURPOSE: Accurate characterization of carotid plaque components is essential for stroke risk assessment. Intraplaque hemorrhage (IPH) is a well-established marker of plaque vulnerability, whereas lipid-rich necrotic cores (LRNCs) often show overlapping signal characteristics with IPH on conventional MRI. The purpose of this study was to compare quantitative susceptibility mapping (QSM) with conventional MRI in differentiating IPH and lipid components in carotid plaques using ex vivo histologic validation. MATERIALS AND METHODS: Twenty-four carotid plaque specimens obtained by en bloc excision during carotid endarterectomy between October 2017 and August 2021 were examined. All specimens underwent ex vivo MRI at 3T, including T1WI, T2WI, and TOF-equivalent FLASH, as well as QSM reconstructed from multi-echo FLASH phase data. After MRI, specimens were serially sectioned and stained using H&E and immunohistochemical techniques. ROIs corresponding to histologically confirmed IPH and lipid components were delineated on MRI by consensus of two radiologists. On conventional MRI, signal intensities were measured in corresponding histologic regions, and contrast-to-noise ratios (CNRs) were calculated relative to surrounding agarose gel. On QSM, absolute magnetic susceptibility values were measured in the same regions. The difference between plaque tissue and the surrounding agarose gel was defined as the relative susceptibility value (rSV), which was used for statistical analysis. CNRs and rSVs among plaque components were compared using the Kruskal-Wallis test, followed by the Steel-Dwass test for pairwise comparisons. RESULTS: Seventy-nine matched cross-sectional slices were analyzed, yielding 24 IPH, 61 LRNCs, and 13 foam-cell aggregates (FAs). No significant difference in CNR was observed between IPH and LRNC on T1WI or FLASH. Histopathologic analysis demonstrated that many LRNCs exhibited erythrocyte-derived remnants consistent with prior hemorrhagic events. FAs showed lower CNRs than IPH on T1WI and FLASH (P < .05). On T2WI, IPH showed lower CNRs than LRNCs or FAs (P < .05). On QSM, IPH demonstrated higher rSVs than LRNCs or FAs (P < .01). CONCLUSION: Histologic validation demonstrates that many LRNCs contain erythrocyte remnants, which may contribute to their overlapping signal characteristics with IPH on conventional MRI. QSM captures erythrocyte-derived susceptibility effects within plaques and provides complementary susceptibility-based information not available from conventional MRI.