2026 Volume 33 Issue 7 Pages 871-881
Intraplaque hemorrhage (IPH) is a key feature of plaque vulnerability that contributes to atherothrombotic events. Non-invasive coronary plaque imaging has been challenging because of the small size of the coronary arteries and motion caused by cardiac contraction and respiration. Recent advances in magnetic resonance imaging (MRI) have enabled the non-invasive detection of coronary IPH. Compared with coronary computed tomography angiography and intravascular imaging modalities, MRI offers unique noninvasive tissue characterization based on intrinsic signal properties. Histopathological and intravascular imaging investigations have indicated that erythrocyte-derived materials, rather than lipid components, constitute the predominant substrate of coronary high-intensity plaques, reflecting recent IPH. This review summarizes the pathophysiological basis, imaging characteristics, and clinical implications of MRI-detected coronary IPH, in the context of other imaging modalities.
Intraplaque hemorrhage (IPH) is recognized as a high-risk feature of atherosclerotic plaques that are associated with atherothrombotic events1). Early insights into IPH were primarily derived from studies of the carotid artery2, 3). This is because surgical specimens are readily available and non-invasive plaque imaging is feasible in large and relatively static carotid arteries. In contrast, non-invasive coronary plaque imaging has been more challenging because of the small size of the coronary arteries and motion caused by cardiac contraction and respiration. The absence of any established diagnostic methods and difficulty in obtaining tissue specimens have limited the clinical investigation of coronary IPH.
Recent advances in imaging techniques are beginning to overcome these challenges, enabling the in vivo assessment of coronary IPH and elucidation of its clinical implications4). This review highlights the recent imaging-based studies of coronary IPH, particularly those using magnetic resonance imaging (MRI), and summarizes the current understanding of its clinical significance.
Two main pathological mechanisms contribute to the formation of IPH5-7). The first is the rupture of immature neovessels, originating predominantly from the adventitial vasa vasorum. Because of structural defects, such as incomplete endothelial junctions, these neovessels are fragile and allow extravasation of blood components into the plaque. The second mechanism involves the incorporation of erythrocytes into plaques. This occurs when luminal thrombi are rich in red blood cells, often developing on ulcerated plaques, undergo healing, and become integrated into the arterial wall.
IPH not only increases the plaque volume, but it also destabilizes plaques. Free cholesterol derived from lipid-rich erythrocyte membranes contributes to cholesterol crystal formation5, 6, 8). In parallel, iron released from hemoglobin induces oxidative stress and initiates inflammatory responses5, 9, 10). Macrophages recruited to clear hemorrhagic debris, including hemoglobin, secrete pro-atherogenic factors, such as vascular endothelial growth factor. This increases the vascular permeability and promotes the development and leakage of immature neovessels. Collectively, these processes lead to the expansion of the necrotic core, amplified inflammation, and loss of fibrous cap integrity, thereby fostering a vulnerable plaque phenotype and increasing rupture risk.
Early, non-invasive identification of IPH is crucial for the risk stratification of atherosclerotic plaques and timely initiation of preventive therapies. Various imaging modalities have been proposed to detect IPH1, 11). A low-intensity area without attenuation on optical coherence tomography (OCT) and an echolucent zone on intravascular ultrasound (IVUS) may indicate histological IPH in vivo12, 13). However, intravascular imaging techniques are generally limited to patients undergoing percutaneous coronary intervention (PCI) and carry the risk of coronary complications. As detailed in the following section, MRI is currently considered best suited for the detection of coronary as well as carotid IPH because its signal characteristics can be delineated using various pulse sequences2, 4, 11).
MRI 1. Carotid IPH DetectionRecent IPH, characterized by a short T1 relaxation time, appears as high signal intensity on T1-weighted images, depending on the oxidative state of hemoglobin11). The three-dimensional magnetization-prepared T1-weighted gradient echo (MP-GRE) sequence is particularly effective for detecting carotid IPH14). By suppressing background signals, it enhances IPH contrast and achieves superior diagnostic accuracy—80% sensitivity and 97% specificity—compared with fast spin-echo and time-of-flight sequences15). MP-GRE also improves the differentiation of IPH from lipid-rich necrotic cores, which often coexist and appear hyperintense on conventional T1-weighted imaging7). Histological validation further confirmed that MP-GRE-positive regions correspond to IPH, but not to lipid-rich necrotic cores in the absence of IPH16). Thus, MP-GRE MRI has been widely adopted as the standard imaging technique for diagnosing carotid IPH11, 14).
2. Evolution and Limitations of Conventional Coronary Plaque MRICoronary plaque MRI remains technically challenging because of the artifacts caused by cardiac and respiratory motion, limited spatial resolution, and prolonged scan times. Nevertheless, technical advancements have gradually, albeit slowly, enabled clinical application. Fayad et al. first demonstrated the feasibility of in vivo coronary plaque imaging using a black-blood MRI technique in humans17). Subsequent refinements by Botnar et al. further improved this technique by integrating black-blood fast spin-echo imaging with real-time respiratory navigators, allowing high-resolution coronary plaque imaging during free breathing18). In 2009, Kawasaki et al. identified coronary high-intensity plaques (HIPs) using a non-contrast, three-dimensional inversion-recovery gradient-echo sequence with respiratory navigator gating on a clinical 1.5-T scanner19). They demonstrated that HIPs were associated with high-risk features on CCTA and IVUS. Since then, the MR sequence has become the most widely applied method for coronary plaque evaluation in clinical research20-25).
3. Recent Advances in Coronary Plaque MRISeveral intrinsic limitations of this sequence have continued to impede its broader clinical adoption26). Owing to the heavily suppressed background in T1-weighted dark-blood images, anatomical reference images, such as bright-blood coronary MR angiography or CCTA, are required to localize target plaques. The separate acquisition of MR angiography not only prolongs the overall scan time, but it also introduces the risk of misregistration between the two datasets, particularly when different MR pulse sequences or spatial and temporal resolutions are employed. The use of CCTA as an anatomical reference adds complexity. Image registration is affected by opposing respiratory phases: end-expiration for dark-blood coronary plaque MRI, and end-inspiration for CCTA.
Novel research sequences allow the simultaneous acquisition of bright- and dark-blood images, mitigating the misregistration problem. Xie et al. developed the Coronary Atherosclerosis T1-weighted CHaracterization with integrated anatomical reference (CATCH) technique, which is based on a three-dimensional, prospective electrocardiogram-gated, inversion recovery-prepared spoiled gradient-echo sequence27). During a single free-breathing three-dimensional whole-heart examination, dark-blood imaging using an inversion recovery pulse is interleaved with bright-blood imaging using a T2-preparation pulse every other heartbeat, with all other MRI parameters kept identical. This provides two spatially matched datasets, dark-blood plaque images and bright-blood coronary artery images, allowing the generation of coregistered fusion images without any image registration steps (Movie 1). Our group previously demonstrated the clinical utility of CATCH MRI28-31), with the selected findings introduced in a later section.
More recently, Ginami et al. introduced the Bright-blood and black-blOOd phase-SensiTive (BOOST) technique32). This method acquires two distinct bright-blood datasets and subtracts them using a phase-sensitive inversion recovery–like reconstruction to generate a third, co-registered black-blood dataset. In a recent report involving a small number of patients with a non-ST-segment elevation myocardial infarction, HIPs on dark-blood images acquired using the BOOST technique were more likely to be associated with the culprit segments. Segments containing lipid-rich plaques, fibroatheromas, thin-cap fibroatheromas, acute plaque ruptures, and intraluminal thrombi showed higher signal intensities than other segments33). Since the black-blood images on BOOST MRI are reconstructed from two bright-blood datasets, it remains unclear whether the clinical interpretations established with other T1-weighted plaque imaging can be directly applied to BOOST MRI. Furthermore, the clinical data of patients with chronic coronary syndrome have not yet been reported. Therefore, the clinical significance of BOOST warrants further investigation.
4. Substrate of Coronary HIP on T1-weighted MRILipid and methemoglobin from recent IPH or thrombus, both of which exhibit short T1 relaxation times, can generate high-intensity signals on T1-weighted MRI4, 34). The frequent coexistence of IPH within lipid-rich plaques complicates the precise understanding of the pathophysiological substrate of coronary HIPs. Ehara et al. reported a higher incidence of intracoronary thrombi on OCT in HIPs, despite no significant difference in the prevalence of lipid-rich plaques between HIP and non-HIP lesions20). This finding may have been influenced by the predominance of acute coronary syndrome in their cohort. In contrast, in an OCT study limited to patients with chronic coronary syndrome, Kanaya et al. observed more lipid-rich plaques and wider lipid arcs in HIPs35). Consequently, coronary HIPs have often been classified under the vague term “vulnerable” or “high-risk” plaques, or alternatively interpreted as lipid-rich plaques34, 36).
Near-infrared spectroscopy (NIRS) enables the quantitative assessment of the lipid components within atherosclerotic plaques, irrespective of the presence of IPH or thrombus13, 37-39). To clarify the substrate of HIPs, our group analyzed 205 coronary plaques in chronic coronary syndrome by comparing CATCH MRI scans with findings on NIRS-IVUS images28). In a multivariable analysis, the maximum 4-mm lipid core burden index was not independently associated with the presence of HIPs. This finding suggests that although lipid-rich components may coexist, erythrocyte-derived materials related to intraplaque hemorrhage appear to play a more prominent role in generating high-intensity signals on T1-weighted coronary MRI.
This interpretation is also supported by histopathological evidence. In an autopsy study, Kuroiwa et al. demonstrated the presence of erythrocyte accumulation and fibrin deposition in HIPs, confirming their association with intraplaque hemorrhage23). Directional coronary atherectomy studies similarly showed that HIP lesions contained greater amounts of erythrocyte-derived materials, fibrin, and iron than non-HIP lesions40, 41). Across these histopathological studies, PMR was positively correlated with levels of immunohistochemical expression of glycophorin A (an erythrocyte membrane protein), fibrin, or tissue factor (a key initiator of blood coagulation)23, 40, 41). A PMR cutoff value of 1.2 was proposed for identifying significant IPH (defined as a glycophorin A-positive area occupying ≥ 5% of the plaque)41).
Taking these findings from NIRS-IVUS and histopathological studies23, 28, 40, 41), MRI serves as a marker of biologically vulnerable plaques, offering a perspective that is complementary to lipid assessment.
Computed Tomography (CT)CT can visualize atherosclerotic plaques and classify them as calcified, fibrous, or soft tissue according to Hounsfield units42, 43). In the carotid, IPH has been reported to show low CT attenuation44). Certain CT features, such as the “rim sign” (adventitial calcification with an internal soft plaque) and increased soft-plaque thickness, have been proposed to improve diagnostic accuracy45). Although both CT and MRI demonstrate comparable diagnostic accuracy (86%) for detecting carotid IPH when validated against histology, MRI is superior for overall IPH assessment46). CT attenuation values, measured in Hounsfield units, show substantial overlap between IPH and other soft-tissue components, such as fibrous tissue and the lipid-rich necrotic core2, 44). In contrast, MRI can differentiate the age of IPH and allow comprehensive assessment of all plaque features45).
Similar to carotid IPH detection, coronary CT angiography (CCTA) has inherent limitations in identifying IPH in the coronary arteries47). Coronary IPH typically appears as a low-attenuation plaque (LAP); however, other plaque components, including a lipid-rich necrotic core, cholesterol clefts, and calcification-related artifacts, may also present as LAP on CCTA48-52).
HIPs are generally assessed using a semi-quantitative index, the plaque-to-myocardium signal intensity ratio (PMR), which is calculated by dividing the highest signal intensity (SI) within the plaque by the mean SI of the myocardium19). As shown in Fig.1, the region of interest for the myocardium is typically selected adjacent to the target plaque at a comparable distance from the chest wall. One report calculated the PMR using the mean plaque SI rather than the highest value as the numerator33). The mean SI of the plaque can vary depending on the selected region of interest, raising concerns about reproducibility. Moreover, this definition yields lower PMR values and therefore requires caution in interpretation. Currently, the original definition of PMR—the highest plaque SI divided by the mean myocardial SI—should be used, as it has been validated against histopathology23) and intravascular imaging28, 35) and has been widely adopted in clinical research21, 25, 28-31, 53).
An example of a PMR calculation is presented. In the upper-left image, a high-intensity signal area is observed in the mid–left anterior descending artery (arrow). Because the myocardium was not visible on the same slice as the plaque, the axial slice was moved slightly caudally to visualize the myocardium at a similar distance from the chest wall (upper-right image). Whole-heart images are shown in Supplementary Video 1. Within the region of interest for the high-intensity plaque, the highest signal intensity (SI) was 179.0 (lower left image). As shown in the lower right image, a region of interest for the myocardium was placed equidistant from the chest wall (coil), yielding a mean myocardial SI of 86.7. The PMR was calculated as the highest plaque SI divided by the mean myocardial SI, resulting in a PMR of 2.06.
PMR: plaque-to-myocardium signal intensity ratio; SI: signal intensity.
The assessment of HIPs using PMR on conventional MR images has an inherent limitation. Typically, a region of interest for the target plaque is manually placed on dark-blood T1-weighted plaque images with a heavily suppressed background by comparing them with the separately acquired anatomical reference images. As mentioned in the previous section, this reliance on visual inspection without the aid of spatially co-registered fusion images introduces ambiguity in the image interpretation. It also requires specialized expertise and a prolonged post-processing time, particularly in cases of long lesions or complex coronary anatomy. This challenge can be addressed using newly developed techniques such as CATCH and BOOST.
The binary categorization of HIP based solely on the highest single-pixel SI does not provide information regarding its spatial extent. To address this, Hosoda et al. introduced a three-dimensional HIP quantification method using a region-growing technique, in which the volume of interest is expanded from a seed point based on visually identified high-intensity areas in dark-blood images. As with the PMR assessment, the absence of spatially co-registered fusion images renders the process dependent on visual interpretation. This increases inter-reader variability, particularly in cases involving long lesions or a complex coronary anatomy. We recently developed a rapid HIP quantification method using CATCH images in combination with a customized fusion software program. As illustrated in Fig.2 (reproduced from Nakazawa M, Matsumoto H, et al., J Cardiovasc Magn Reson 2024; 26:100999, under CC BY-NC-ND 4.0 license)30), the software program automatically generates tubular three-dimensional volumes of interest on dark-blood plaque images along the coronary vasculature, guided by precisely co-registered bright-blood coronary artery images, and subsequently calculates both the PMR and HIP volumes. Supplementary Video 2, available on the journal website, presents synchronized whole-heart axial dark-blood, bright-blood, and fused images.
Representative workflow for semi-automated HIP quantification using Coronary Atherosclerosis T1-weighted CHaracterization (CATCH) images. Once the lumen centerline is manually traced along the target segment on bright-blood coronary artery images, the software automatically generates tubular three-dimensional volumes of interest encompassing the coronary artery and its surroundings (left panels). Color-coded overlays in the right panels correspond to the signal intensity on the inherently co-registered dark-blood T1-weighted images. Note that the volumes of interest cover high-intensity plaques, indicated by orange regions. The signal intensity of each voxel within the designated volume of interest is computed to determine the volume of voxels exceeding the predefined signal intensity threshold (i.e., HIP volume). Synchronized whole-heart axial dark-blood, bright-blood, and fusion images are presented in Supplementary Video 2. The bright and dark blood images were spatially matched without manual registration.
Reproduced by Nakazawa M, Matsumoto H, et al. J Cardiovasc Magn Reson 2024;26:100999 under the CC BY-NC-ND 4.0 license.
CATCH: Coronary Atherosclerosis T1-weighted Characterization, HIP: high-intensity plaque.
The ISCHEMIA trial reported no overall reduction in ischemic cardiovascular events with an initial invasive strategy in patients with chronic coronary syndrome. This outcome may be partly attributable to the relatively low cardiovascular event rate (18.2% at 5 years) observed in the initial conservative strategy of medical therapy alone54). In contrast, the PREVENT trial demonstrated the benefit of preventive PCI targeting non-flow-limiting yet vulnerable plaques identified by intravascular imaging55). However, in routine clinical practice, intravascular imaging is generally limited to patients undergoing PCI. This underscores the need for noninvasive approaches to identify high-risk plaques.
The prognostic value of MRI-detected IPH is particularly well established for the carotid arteries1, 11). Meta-analyses have confirmed that MRI-detected carotid IPH is a strong predictor of future ischemic cerebrovascular events1, 11), independent of plaque thickness or traditional risk factors56). Owing to the limited adoption of coronary plaque MRI, clinical evidence regarding its prognostic value is limited to a single-center study by Noguchi et al.21) In that study, non-contrast T1-weighted MRI was performed in 568 patients with coronary artery disease initially screened by CCTA. Over a median follow-up period of 55 months, cardiac events occurred in 55 patients (9.7%), including cardiac death, acute coronary syndrome, and ischemia-driven PCI due to progressive angina pectoris. A receiver-operating characteristic curve analysis identified a PMR of 1.4 as the optimal cutoff value for predicting cardiac events. Cardiac events occurred in 25.8% of patients with PMR ≥ 1.4, 8.4% with PMR 1.0–1.4, and 1.1% with PMR ＜1.0. Notably, half of the coronary events arising from segments with a PMR ≥ 1.4 occurred within 12 months. A multivariable analysis identified PMR ≥ 1.4 as the strongest independent predictor of coronary events, with a hazard ratio of 3.96.
In line with the previous report, we recently encountered a patient with a HIP on CATCH MRI who developed an ST-segment elevation myocardial infarction within a few months of imaging (the same case is shown in Fig.1 and Supplementary Video 1; available on the journal website). Among several atherosclerotic lesions with mild to moderate stenosis, emergency coronary angiography revealed that the moderately stenotic lesion harboring the HIP had progressed to total occlusion. The details of this case are shown in Fig.3. This case underscores the potential of T1-weighted MRI, including CATCH, to noninvasively identify biologically active coronary plaques directly linked to acute plaque destabilization.
On a curved planar reformatted coronary computed tomography (CT) angiogram (A), three atherosclerotic lesions with mild-to-moderate stenosis and low-attenuation plaques were observed from the left main trunk to the mid–left anterior descending artery (LAD). A multiplanar reconstruction fusion CATCH magnetic resonance imaging (MRI) image demonstrated a high-intensity plaque with a PMR of 2.06 in the third lesion of the mid-LAD (B). Synchronized whole-heart axial CATCH images are shown in Supplementary Video 1. An ST-segment elevation myocardial infarction occurred four months after CT and MR imaging. Emergency invasive coronary angiography revealed a total occlusion of the mid-LAD (C). Notably, the site of total occlusion corresponded to the third lesion with the HIP on CATCH MRI.
CATCH, Coronary Atherosclerosis T1-weighted Characterization, HIP, high-intensity plaque; CT, computed tomography; LAD, left anterior descending artery; MRI, magnetic resonance imaging; PMR, plaque-to-myocardium signal intensity ratio; SI, signal intensity.
The PROSPECT II trial, a multicenter prospective natural history study using NIRS-IVUS, characterized 3,629 untreated non-culprit lesions in 898 patients with recent myocardial infarction following successful treatment of culprit lesions57). Over a median follow-up of 3.7 years, lipid-rich plaques, defined by a maximum 4-mm lipid core burden index ≥ 324.7, independently predicted major adverse cardiac events arising from untreated non-culprit lesions. These events included cardiac death, myocardial infarction, unstable angina, and progressive angina. Nevertheless, the 4-year lesion-level event rate for lipid-rich plaques was approximately 4%, corresponding to a consistent annual risk of 1%.
In contrast, lesions with a low-intensity area on OCT, particularly when co-localized with cholesterol crystals58), and those with an echolucent zone on IVUS59) were associated with early clinical events within 6–12 months, followed by a more gradual increase over time. Recent studies have highlighted the potential association between IPH and coronary spasm as a mechanism contributing to rapid plaque progression and sudden cardiac arrest. Nishi et al. reported that IPH, identified by OCT, was frequently observed in patients resuscitated from sudden cardiac arrest without obstructive coronary lesions and was often accompanied by coronary spasm demonstrated by acetylcholine provocation testing60). Although the causal relationship between IPH and coronary spasm remains incompletely understood, their interaction may, in part, elucidate important clinical implications of IPH.
High-risk plaque characteristics on CCTA, including LAP, positive remodeling, and the napkin-ring sign, have been reported to be associated with coronary events61). Among these, LAP volume or burden has shown the strongest predictive value62, 63). Interestingly, Kaplan–Meier curves from a study linking LAP burden to coronary events62) revealed a pattern similar to that reported by Noguchi et al. for MRI-detected HIPs21). In both studies, events clustered relatively early, followed by a more gradual slope. Based on the findings of MRI and NIRS studies, early coronary events arising from LAPs may, in part, reflect underlying IPH. Recent guidelines advocate CCTA as the preferred first-line test for coronary artery disease64, 65). Considering the non-specific nature of LAP, which includes IPH, other plaque components, and imaging artifacts48-52), the selective addition of MRI in patients with LAP may be clinically feasible and could enhance risk stratification.
Pathological and intravascular imaging studies have demonstrated that erythrocyte-derived components (i.e., recent IPH) constitute the predominant substrate of coronary HIPs identified using non-contrast T1-weighted MRI in chronic coronary syndrome. MRI-detected IPH appears to signify more imminent coronary events, shifting the paradigm from lipid-rich plaques, which are long regarded as the hallmark of vulnerability. Continued technical advancements in MRI may enable the detection of truly high-risk coronary plaques in routine clinical practice, without radiation exposure or the need for contrast agents.
The author is grateful to Drs. Yibin Xie and Debiao Li at the Biomedical Imaging Research Institute, Cedars-Sinai, for providing the CATCH technique to our institution. The author also wishes to thank Drs. Damini Dey and Piotr J. Slomka at the same institute for software support used for quantitative analysis of HIPs.
None.
