Multimodal Imaging Enables Identification of Regional and Temporal Structural Remodeling Following Myocardial Infarction

Cardiac rupture is a rare but often fatal complication of myocardial infarction (MI)[1,2]. Late reperfusion therapy (LRT), restoring flow after cardiomyocyte death, reduces rupture incidence[1,3]. However, it does not reduce infarct size or transmurality [4,5]. Given the localized nature of rupture, regional changes in tissue content and structure are potential contributors to rupture events [6,7]. Therefore, in this study we quantified the temporal and regional changes in structural heterogeneity following MI in rats. We applied multimodal microscopy and algorithm image segmentation to visualize variations in the ventricular wall up to five days post MI. Male Sprague-Dawley rats (n=31) were subjected to MI by either permanent (PO) or temporary (LRT, 3 hours of occlusion) occlusion of the coronary artery. Hearts were harvested on days 1, 3, and 5 post-MI. Experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals and were reviewed and approved by the University of Wisconsin IACUC. Histological sections were stained with hematoxylin-eosin (H&E), Picrosirius Red (PSR) or Masson’s Trichrome (Tri). Samples were imaged using brightfield, standard linearly polarized light, and/or label-free liquid-crystal-based polarized light (PolScope) microscopy [8]. We validated PolScope imaging of neighboring H&E-stained sections with brightfield images of Tri-stained samples. Infarct visualization was comparable on day 5, when necrotic myocytes had been cleared. However, PolScope provided superior visualization at post MI days 1 and 3. Therefore we used PolScope for infarct identification and spatial segmentation. A k-means segmentation algorithm was applied to the PolScope images of H&E-stained sections to identify infarct core, infarct border, and peripheral regions and determine their relative sizes. Brightfield images of H&E-stained sections and standard linearly polarized light images of PSR stained sections were processed using custom MATLAB code to isolate cell nuclei and collagen fibers, respectively. Region masks from PolScope were then mapped to these images to quantify regional cellular and collagen densities. Two-way ANOVAs were then conducted with post hoc multiple comparison testing completed via Tukey testing. Border thickness decreased over time following MI and this trend progressed more quickly in the LRT group. Cell density in the infarct core region was lower on day 1 than days 3 and 5, suggesting the immune response intensified with time or that, as large necrotic myocytes were cleared, there was an increase in spatial cell density. There were no significant differences in cell density between groups. On day 1, collagen density was significantly higher in the infarct border and core regions of PO samples than LRT samples. This trend reversed on day 3, with collagen density of LRT samples exceeding that of PO samples. At days 3 and 5, core collagen density was significantly higher than peripheral collagen density for LRT samples and at day 5 this trend also appeared in the PO group. These regional differences could suggest that early remodeling during the necrotic phase of healing − characterized by breakdown of the collagenous ECM − occurred more quickly in LRT. We conducted a quantitative analysis comparing regional and temporal differences in early structural remodeling following MI using multimodal imaging. We found the combination of brightfield, polarized light and PolScope microscopy highly effective at identifying temporal and spatial changes post MI in cellular and collagen content. In particular, PolScope was instrumental in automated segmentation of infarct regions during the necrotic phase of healing, especially prior to substantial clearing of damaged and necrotic myocytes. Our findings suggest that LRT accelerates the healing process, reducing the duration during which the LV wall is susceptible to rupture events [1,9].