Shear Wave Elastography Can Differentiate between Radiation-Responsive and Non-responsive Pancreatic Tumors: An ex Vivo Study with Murine Models

Abstract

Neither contrast-enhanced computed tomography nor magnetic resonance imaging can monitor changes in the pancreatic ductal adenocarcinoma microenvironment during therapy. We hypothesized that shear wave elastography could overcome this limitation. To test this hypothesis, we measured the shear modulus of two groups of murine pancreatic tumors (KCKO, n = 30; PAN02, n = 30) treated with stereotactic body radiation therapy (SBRT). The mean shear modulus of KCKO tumors was 7.651 kPa higher than that of PAN02 tumors (p < 0.001). SBRT reduced the shear modulus in KCKO tumors by 8.914 kPa (p < 0.001). No significant difference in the shear modulus of SBRT-treated PAN02 tumors was observed. Additionally, necrotic and collagen densities were reduced only in the SBRT-treated KCKO tumors. Shear modulus was dependent on collagen distribution and histological texture parameters (i.e., entropy and fractional dimension). Shear wave elastography imaging differentiates between SBRT-responsive (KCKO) and non-responsive (PAN02) tumors.

Keywords: Pancreatic ductal adenocarcinoma; Shear modulus; Stereotactic body radiation therapy; Tumor microenvironment.

Figure 1.

Experimental setup and examples of tumor shear modulus, BLI tumor growth, and Masson’s trichrome images for representative KCKO and PAN02 murine pancreatic cancer tumors. (a) Verasonics Vantage 256 ultrasound system equipped with an L-7 transducer probe, positioned on top of a tumor enclosing phantom where the transducer scan plane was marked with a dotted blue line. (b) In vivo imaging system (PerkinElmer) and representative BLI results for monitoring the tumor growth burden in mice with pancreatic cancer tumors. (c) and (d) are the ultrasound sonogram and shear modulus elastogram for two representative KCKO and PAN02 murine pancreatic cancer tumors, respectively. The (e) KCKO and (f) PAN02 tumors were sectioned along the scan plane and stained for Masson’s Trichrome for collagen content analysis.

Figure 2.

Illustration for collagen entropy calculation. (a) Segmented collagen fibers from the Masson’s trichrome-stained images. The segmented collagen was split into 100 μm by 100 μm windows and the collagen fractional area density was calculated and assembled from each window and shown in (b). (c) Histogram of the collagen fractional area density (%) from which the collagen entropy was calculated.

Figure 3.

The box-cutting method for collagen fractional dimension (D) calculation. (a) to (c) illustrate the box-counting method where N is the number of boxes per dimension. (d) Log (box counts) vs. log (number of boxes). The slope of the XY plot represents the fractional dimension and was calculated from least squares estimation method.

Figure 4.

The texture energy measures. (a) Masson’s trichrome-stained tumor image with regions of (b) collagen fibers and (c) necrotic tissues with similar color features. The L5E5/E5L5 texture energy maps for the corresponding (d) collagen fibers and (e) necrotic tissues.

Figure 5.

Bioluminescence and shear modulus for KCKO and PAN02 tumors at animal sacrifice. (a) The geometric-mean-bioluminescence signal (p/sec/cm²/sr) for the control- and SBRT-treated KCKO tumors taken over 24 days in comparison to those taken over 17 days for the PAN02 mice. The shear moduli at day 24 and 17 after animal sacrifice were plotted for a representative tumor from each group. (b) The BLI signal for each KCKO and PAN02 tumor plotted against the corresponding shear modulus.

Figure 6.

The shear modulus images of SBRT treated KCKO and PAN02 tumors. (a) and (b) are shear modulus elastograms overlaid on sonograms obtained from representative KCKO control and SBRT-treated tumors, respectively. (c) and (d) are the shear modulus elastograms overlaid on sonograms obtained from representative PAN02 control and SBRT-treated tumors, respectively. (e) Boxplot of the average shear modulus distribution for each tumor line.

Figure 7.

Necrosis area and collagen density analyses of KCKO and PAN02 tumors. (a) and (b) are boxplots of necrosis density (%) and collagen density (%) for KCKO control, KCKO SBRT-treated, PAN02 control and PAN02 SBRT-treated groups (**** represents p ≤ 0.0001, ** represents 0.001 ≥ p ≤ 0.01, * represents 0.01 ≥ p ≤ 0.05, ns represents p > 0.05). The relationship between average shear modulus and (c) necrosis density for PAN02 tumors, and (d) collagen densities for KCKO and PAN02 tumors.

Figure 8.

Entropy and fractional dimension analyses for the KCKO and PAN02 tumors. (a) and (b) boxplots of collagen entropy and fractional dimension for the KCKO control, KCKO SBRT-treated, PAN02 control, and PAN02 SBRT-treated groups. The relationship between shear modulus and (c) collagen distribution entropy and (d) fractional dimension for both groups of tumors.