Spectral computed tomography may improve assessment of carotid plaques by measuring changes in the fat around affected arteries. The findings, published in Insights into Imaging, come from a retrospective single-centre study of 306 patients with computed tomography angiography-confirmed carotid atherosclerosis. Symptomatic plaques showed different perivascular adipose tissue features and plaque composition from asymptomatic plaques. Adding selected spectral measurements to stenosis severity and necrotic core volume improved separation between the 2 groups. The approach therefore provided information beyond arterial narrowing alone, although the findings still require validation in other centres and patient groups.
Must read: Spectral CT Helps Distinguish Periprosthetic Infection
Spectral Measures Show Clear Tissue Differences
The analysis included 169 patients with symptomatic plaques and 137 with asymptomatic plaques. Symptomatic status meant an ipsilateral ischaemic stroke, transient ischaemic attack or temporary loss of vision in one eye within the previous 6 months. Brain magnetic resonance imaging supported confirmation of the ischaemic lesions and their link with the affected carotid artery. Patients without clinical or imaging evidence of such events during the same period were classified as asymptomatic.
Spectral computed tomography angiography assessed perivascular adipose tissue around the point of greatest arterial narrowing. Measurements included effective atomic number, iodine concentration, fat fraction and attenuation on conventional and virtual monoenergetic images. Plaque burden, length and the volumes of fibrous, fibrous fatty, necrotic and calcified tissue were also measured.
Compared with asymptomatic plaques, symptomatic plaques had more severe stenosis, a higher plaque burden and greater length. They also contained more fibrous fatty tissue and larger necrotic cores, while fibrous tissue volume was lower. Plaque thickness and calcium volume did not differ significantly.
The surrounding fat also showed clear differences. Symptomatic plaques had higher effective atomic number, iodine concentration and attenuation values, together with a lower fat fraction. The slope of the spectral attenuation curve did not differ between groups. Agreement between the 2 radiologists was substantial to excellent for the plaque and perivascular measurements.
Surrounding Fat Reflects Plaque Composition
Several spectral features of perivascular adipose tissue were associated with plaque composition. Higher effective atomic number and iodine concentration were linked with larger fibrous fatty and necrotic core volumes. Higher attenuation on virtual monoenergetic images showed similar positive associations. In contrast, fat fraction fell as necrotic core volume increased. Fibrous volume was negatively associated with several spectral measures, while calcium volume showed no significant association with the perivascular parameters.
These relationships show that spectral imaging captures differences in the tissue around carotid plaques as well as changes within the plaques. However, the findings do not show whether changes in perivascular adipose tissue occur before plaque instability or develop in response to plaque activity, vascular inflammation or ischaemic events. The cross-sectional design therefore does not support conclusions about cause and effect.
Subgroup analysis showed that several perivascular measurements still separated symptomatic from asymptomatic plaques in patients with mild or moderate stenosis. Symptomatic plaques in this group had higher effective atomic number, iodine concentration and attenuation, as well as lower fat fraction. In patients with severe stenosis, symptomatic plaques still showed higher effective atomic number and lower fat fraction, but iodine concentration and conventional attenuation were no longer significantly different.
This pattern suggests that some spectral measurements may remain useful even when arterial narrowing is not severe. The analysis did not assess whether these differences predict future stroke or transient ischaemic attack.
Combined Models Improve Identification
The individual plaque and perivascular measurements varied in their ability to identify symptomatic plaques. Among the spectral measures, attenuation at 70 keV and fat fraction performed best. Effective atomic number also showed useful discrimination, while iodine concentration was less accurate. Among plaque components, fibrous volume and necrotic core volume performed better than fibrous fatty volume.
A model based only on stenosis severity had an area under the curve of 0.716. Adding necrotic core volume increased the value to 0.821. Adding fat fraction raised it further to 0.897, while adding attenuation at 70 keV increased it to 0.916. Both additions significantly improved discrimination beyond the model that included stenosis and necrotic core volume.
After adjustment for clinical characteristics, fibrous volume, fibrous fatty volume, necrotic core volume, effective atomic number, fat fraction, iodine concentration and attenuation at 70 keV remained independently associated with symptomatic plaques. Internal validation also showed that the main associations were generally stable across repeated samples.
Several factors limit clinical use. The analysis was retrospective, involved 1 centre and had no external validation. Plaque status was based on clinical presentation and magnetic resonance imaging rather than tissue examination. Intraplaque haemorrhage was not assessed. The measurement regions had to be placed manually, making the process time-consuming and dependent on the operator. The spectral measures also cannot be obtained from standard single-energy computed tomography angiography.
Spectral computed tomography measurements of perivascular adipose tissue differed between symptomatic and asymptomatic carotid atherosclerosis and were linked with plaque composition. Fat fraction and attenuation at 70 keV added information beyond stenosis severity and necrotic core volume. Assessing the tissue around the artery may therefore complement conventional plaque imaging. However, the single-centre retrospective design, lack of external validation and dependence on spectral imaging and manual measurements limit immediate clinical use. Prospective studies with follow-up are needed to confirm predictive value and wider applicability.
Source: Insights into Imaging
Image Credit: iStock
References:
Zheng X, Liu X, Song J et al. (2026) Spectral CT parameters of perivascular adipose tissue as non-invasive biomarkers for identifying symptomatic carotid atherosclerosis. Insights Imaging; 17, 190.
