The finding suggests that the composition of an atherosclerotic plaque may carry information relevant to the health of the brain’s small vessels. Zeff is generated from spectral CT data as an estimate of tissue composition. In the study, patients with high CSVD burden had a mean Zeff of 7.644, compared with 7.897 among those with low burden. They also had lower attenuation at all three virtual monoenergetic energies, a less steep spectral curve, lower effective electron density and thicker non-calcified plaque. Iodine density, by contrast, did not differ significantly between the groups, with a p value greater than 0.05.

Researchers assessed CSVD with 3.0-tesla brain magnetic resonance imaging and assembled a total score from 0 to 4. One point was assigned for the presence of a lacune, substantial white matter hyperintensity, an enlarged basal-ganglia perivascular-space category, or at least one cerebral microbleed. The MRI examinations occurred within 30 days of carotid CT angiography. Of the 216 participants, 125 had low burden, defined by scores below 2, and 91 had high burden, defined by scores of 2 or more. The distribution was 46 patients at score 0, 79 at score 1, 50 at score 2, 35 at score 3 and 6 at score 4.

The cohort was recruited prospectively for neurological evaluation and included people with suspected carotid disease, recent transient ischemic attack or minor stroke, and other neurological manifestations. Imaging was performed on a Philips Hawk Spectral CT 7,500 scanner. Virtual monoenergetic images at 40, 70 and 140 keV were reconstructed together with Zeff, electron-density and iodine-density maps. For plaque analysis, radiologists placed a region of interest of approximately 1.5 to 3.0 square millimeters on the 40 keV image at the site of most severe stenosis, avoiding calcified tissue and averaging repeated measurements. The region was synchronized automatically across the other reconstructed datasets.

Conventional plaque characteristics were assessed independently by two radiologists with five and eight years of experience, with a third senior radiologist resolving disagreements. These observers were blinded to brain MRI results and clinical information. The DLSCT parameters were also analyzed independently by the two radiologists, but the study does not describe third-radiologist adjudication for those measurements. Conventional features included degree of luminal stenosis, non-calcified plaque thickness and plaque ulceration. Stenosis was calculated with North American Symptomatic Carotid Endarterectomy Trial criteria and classified as mild, below 30%; moderate, 30% to 69%; or severe, 70% to 99%.

Repeat measurements in 50 randomly selected cases were used to evaluate intra-rater and inter-rater reliability. The first radiologist re-measured the same cases after four weeks, while intraclass correlation coefficients and Bland-Altman analysis assessed agreement. Two alternative Zeff strategies were tested: the lowest value among measurable plaques in both carotid arteries and the average across all measurable plaques. Both remained associated with high CSVD burden after adjustment for age, hypertension, diabetes, plaque thickness and carotid stenosis, with odds ratios of 0.210 and 0.230 per one-standard-deviation increase, respectively, and p values below 0.001. These sensitivity analyses supported the result obtained from Zeff measured at the most stenotic plaque.

In multivariable logistic regression, age was associated with higher CSVD burden in every model, while Zeff had a particularly strong inverse association. A one-standard-deviation increase in standardized Zeff corresponded to an odds ratio of 0.387, with a 95% confidence interval of 0.262 to 0.572, for high burden in a model containing age, hypertension and diabetes. When plaque thickness and stenosis were also included, the odds ratio was 0.386, with a 95% confidence interval of 0.256 to 0.581; both p values were below 0.001. By comparison, neither non-calcified plaque thickness nor moderate or severe carotid lumen stenosis remained significantly associated with high burden after adjustment. The result does not establish that Zeff causes small-vessel injury; rather, it identifies a measurable imaging feature correlated with the extent of brain injury in this selected cohort.

The four classification models were compared using optimism-corrected discrimination, calibration, Brier score and decision curve analysis. The clinical model, containing age, hypertension and diabetes, had an AUC of 0.624, with a 95% confidence interval of 0.590 to 0.733. Adding conventional plaque features produced an AUC of 0.639, with a 95% confidence interval of 0.585 to 0.729. Models incorporating Zeff performed better, with AUCs of 0.748, with a 95% confidence interval of 0.712 to 0.837, for the clinical-plus-Zeff model and 0.759, with a 95% confidence interval of 0.708 to 0.835, for the model also including plaque thickness and stenosis. The fully adjusted model had the lowest Brier score, 0.199, and a calibration slope of 0.933, close to the ideal value of 1. The AUC difference between the conventional imaging model and the fully adjusted model was 0.120, with a 95% confidence interval of 0.058 to 0.183 and a paired DeLong test yielding p less than 0.001.

Decision curve analysis indicated that the two models containing Zeff offered greater net benefit across most evaluated threshold probabilities. The largest advantage for the fully adjusted model appeared at thresholds of 20% to 50%, particularly from 30% to 40%, although these were exploratory analyses rather than evidence that the models are ready for clinical deployment. Subgroup tests found no statistically significant interaction between Zeff and examined factors such as age, sex, smoking, hypertension, diabetes, carotid stenosis, prior stroke or transient ischemic attack, acute small infarction and symptomatic carotid disease. This consistency does not substitute for validation in independent populations.

The investigators proposed several mechanisms that could connect plaque composition with CSVD, while emphasizing that these interpretations remain speculative. One possibility is that inflammation within a vulnerable plaque contributes to cerebral microvascular endothelial injury and blood-brain barrier disruption. Another is that embolic material from such plaques obstructs small arterioles, producing focal perfusion disturbances that may appear as white matter lesions, lacunes, enlarged perivascular spaces or microbleeds. A third pathway involves endothelial dysfunction affecting cerebral microcirculation and barrier integrity. The cross-sectional design cannot establish temporal order or causation, and the researchers noted that histopathological confirmation and longitudinal studies would be needed to test these explanations. The lower Zeff values observed in the high-burden group were interpreted as potentially indicating larger proportions of lipid-rich, hemorrhagic or necrotic components, but the imaging measurements were not validated directly against plaque histology.

Several limitations constrain the results. All participants came from one institution and were hospitalized patients undergoing clinically indicated carotid imaging and MRI, so the findings may not apply to asymptomatic people in the wider community. The models received internal bootstrap correction but no external validation, and the sparse distribution of the highest CSVD scores prevented reliable ordinal regression, leading the researchers to dichotomize burden into high and low categories. Plaque composition was represented by imaging surrogates rather than direct histology, and manual region-of-interest placement leaves room for operator variability. The study therefore supports Zeff as a promising, independently associated imaging correlate of CSVD burden, but its diagnostic accuracy, biological interpretation and usefulness beyond the studied hospital population remain to be established in larger, multicenter and prospective cohorts.

Subject of Research: Association between spectral CT-derived carotid plaque effective atomic number and cerebral small vessel disease burden.

Article Title: Quantitative parameters of carotid plaque based on dual-layer spectral-detector CT for stratifying cerebral small vessel disease

Article References: Ye, Y., Lin, L., Liu, Y., Lin, C., & Xu, Y. (2026). Quantitative parameters of carotid plaque based on dual-layer spectral-detector CT for stratifying cerebral small vessel disease. Frontiers in Neurology, 17, Article 1911566. https://doi.org/10.3389/fneur.2026.1911566

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1911566

Keywords: cerebral small vessel disease, carotid atherosclerosis, spectral CT, effective atomic number, carotid plaque, brain MRI, cross-sectional study, medical imaging, plaque composition, diagnostic modeling, Quantitative, parameters