In-vivo carotid T2 mapping can accurately quantify plaque lipid content to discriminate between symptomatic and asymptomatic patients: histological validation, scan-rescan reproducibility and clinical study
Journal of Cardiovascular Magnetic Resonance volume 18, Article number: W10 (2016)
In-vivo carotid CMR is able to identify features of plaque vulnerability such as lipid core size. However, the current standard (multicontrast CMR) requires contrast media, extensive post-processing and subjective interpretation. Recently we proposed to use quantitative T2 mapping to distinguish plaque lipid from surrounding fibrous tissue and measure lipid core size. This study aimed to (1) validate plaque lipid quantification by T2 mapping against histology and (2) investigate if it could discriminate between symptomatic and asymptomatic patients.
40 patients scheduled for carotid endarterectomy (50-99% stenosis on Ultrasound), either symptomatic or asymptomatic, were imaged at 3T (Siemens Verio) max 24 h before surgery (IRB approved, written consent obtained). We used our novel black-blood Multiecho Spin-Echo sequence for T2 mapping (DANTE-MESE) to acquire 5 slices in 4 min (TR = 2 s, TE = 9-18-…-126 ms, partial Fourier = 5/8, FOV = 128 × 128 mm, matrix size = 384 × 384, slice thickness = 2 mm, slice gap = 2 mm). T2 maps were generated using nonlinear fitting; lumen and external carotid boundary were contoured semi-automatically; lipid core was segmented from other plaque components by T2 thresholding; and the optimal segmentation (i.e. highest correlation with histology) was automatically calculated using leave-one-out cross-validation (Figure 1).
Plaques were collected at the time of carotid endarterectomy, processed and cut at 1 mm intervals using carotid bifurcation and T1w images as references to match T2 map locations. Plaque lipid area was manually segmented on sections stained with H&E and Masson Trichrome, with Oil-red-O used to confirm lipid distribution.
26 patient scans (median age 69) had good quality and 60 slices with matching locations on plaque histology. Lipid without haemorrhage had shorter T2 than normal vessel wall and fibrous tissue, whereas haemorrhage infiltrating the lipid core increased T2. Thresholding T2 < 42 ms and T2 > 90 ms resulted in the globally optimal lipid core segmentation (highest R = 0.85 with histology, Figure 2A). We found significantly more lipid in symptomatic than in asymptomatic plaques (Figure 2B); ROC analysis showed a fair/good ability to discriminate between the 2 groups; and the optimal cut-off value ~25% agreed with histological studies that found higher risk of rupture associated with lipid core >25% of plaque area (Figure 2C). Scan-rescan reproducibility in 9 patients was excellent: ICC = 0.89 (95% CI 0.59-0.98) and CoV = 8.9%.
We have demonstrated that carotid T2 mapping can be used to quantify plaque lipid content with good accuracy and reproducibility, and to classify plaques as symptomatic or asymptomatic based on their lipid core size. This technique can potentially be used to identify patients at risk of plaque rupture; informing decisions of stents vs. surgery; stratify for more intensive lipid treatment; and monitor response to treatment in clinical trials.
About this article
Cite this article
Biasiolli, L., Chai, J.T., Li, L. et al. In-vivo carotid T2 mapping can accurately quantify plaque lipid content to discriminate between symptomatic and asymptomatic patients: histological validation, scan-rescan reproducibility and clinical study. J Cardiovasc Magn Reson 18 (Suppl 1), W10 (2016). https://doi.org/10.1186/1532-429X-18-S1-W10
- Carotid Endarterectomy
- Lipid Core
- Plaque Component
- Plaque Vulnerability
- Masson Trichrome