Identification of atherosclerotic lipid deposits by diffusion-weighted imaging

Abstract

Objectives: The content and distribution of lipids is an important aspect of plaque vulnerability, but lipids are present within a heterogeneous environment, impeding detection by magnetic resonance imaging. Our goal was to achieve accurate detection of mobile lipids by a single magnetic resonance imaging sequence.

Methods and results: Carotid endarectomy specimens (n=23) were imaged ex vivo at a high magnetic field (11.7 T) within 24 hours after surgery. Three contrast-weighted (T1W, T2W, and diffusion-weighted imaging [DWI]) image sequences were acquired and then coregistered with histological preparations for lipids (Oil red O and polarized light microscopy) and fibrous tissue (trichrome). Contrast-to-noise ratios were measured and compared for the 3 contrast weightings. Contrast-to-noise ratio measurement in regions identified as lipid versus fibrous tissue showed greater differences by DWI (4.5+/-0.63 versus 0.64+/-0.08; P<0.05) as compared with T2W (2.83+/-0.36 versus 1.36+/-0.37; P<0.05). We validated the presence and distribution of lipids (mainly cholesteryl esters) by both histology and image-guide spectroscopy. The basis for distinguishing mobile lipid and water inside the plaque was illustrated by diffusion-weighted spectroscopy.

Conclusions: Biophysical properties of plaque lipids can confer selective identification by DWI, as opposed to standard T1W and T2W imaging sequences. Successful translation of DWI in vivo could identify of features of vulnerable plaque.

Figure 1

T1W and T2W MR images of 2 carotid plaques (endarectomy specimens) and histological identification of lipids. A and D, T1W image. B and E, T2W image. C and F, ORO staining: intense red staining signifies lipid deposits.

Figure 2

T1W, T2W, and DW MR images of a carotid plaque (endarectomy specimen) correlated with histology. A, T1W image. B, T2W image. C, DW image (b value 1230 s/mm²). D, ADC map. E, ORO staining. Major features of the plaque are indicated. F, Polarized light microscopy image of relevant unstained section at 22°C. Bright signals indicate lipids (mainly CE). The inset shows Malt-ese crosses characteristic of liquid-crystalline CE after heating to 60°C and cooling to 33°C.

Figure 3

A, CNR for different tissue components in T1W, T2W, and DW images. CNR values are shown as mean±SEM, N = 17. In the plot, zero represents the contrast value of background PBS. *Statistically significant difference. B, Histograms for 3 parametric maps. The horizontal axis is measured T1, T2, and ADC values; vertical axis is number of pixels per unit. The bin sizes for the x-axis are 0.02×10⁻³ mm²/s for ADC, 25 ms for T1 map, and 1 ms for T2 map, respectively. The locations of histogram peaks with respective to x-axis indicate most common ADC, T1, and T2 values.

Figure 4

Image-guided proton spectra of a carotid plaque correlated with histology. A, Spectra obtained from lipid-rich and lipid-poor regions, respectively, with water suppression. B, Spectrum obtained from a lipid mixture (1:1 cholesteryl linoleate/cholesteryl oleate mixture). C, T2W image. D, DW image. E, ORO staining.

Figure 5

A and B, Plot of normalized signal intensity versus b factors (attenuator factor) in high b value range (0 to 16 000 s/mm²) and a low b range (0 to 700 s/mm²) in pure lipid and water phantoms. Water signal does not go to zero because measurements were performed on magnitude images where the noise does not average to zero. C, Serial spectra obtained from the lipid-rich regions by varying the b value (same voxel as shown in the Figure 4). D, Plot of normalized signal intensity versus b value in the lipid-rich region of the plaque.