INTRODUCTION

Real MR elastography (r-MRE) is used much less in the rectum than in the liver probably due to the external elastography equipment^1,2. To solve this dilemmachallenge, some studies suggested that Intravoxel Incoherent Motion imaging (IVIM)-based virtual elastography (v-MRE) could been used as an alternative to r-MRE for the staging assessment of liver fibrosis, using the relationship between shifted apparent diffusion coefficient (sADC) and tissue stiffness^3,4. However, the existence of a relationship between water diffusivity and tissue stiffness remains unclear in rectal adenocarcinoma (RA). Our aim was to (1) explore the optimal combination of b-values for v-MRE based on IVIM images in RA tissue; (2) investigate the feasibility of v-MRE based on IVIM images for preoperatively assessing the tumor (T), node (N) and prognostic (P) stage in RA with comparison of r-MRE.

METHODS

Patient recruitment: The study was approved by local institutional ethics committee, and written informed consent was obtained from all patients. A total of 50 RA patients (mean age, 61 years ± 10 [standard deviation], 58% males) were prospectively enrolled with IVIM and r-MRE scans before surgery.
MRI study: All images were acquired on a 3T MRI scanner (MAGNETOM Prisma, Siemens Healthineers, Erlangen, Germany). IVIM images were acquired with a single-shot echoplanar imaging (TR = 3700.0 ms; TE = 50.0 ms; FOV = 380×380 mm; image matrix = 164×164; voxel size = 1.2×1.2×3.0 mm³; b = 0, 30, 60, 100, 150, 200, 500, 1000, 1500 and 2000 sec/mm²; acquisition time = 5 min). r-MRE images were acquired with a 3D single-shot spin-echo echoplanar imaging along three motion-encoding directions (TR = 1670.0 ms; TE =56.0 ms; FOV = 300×300 mm; image matrix = 100×100; voxel size =3.0×3.0×3.0 mm³; acquisition time = 3 min 7 sec). v-MRE map was generated through research post-processing software using IVIM images.
Histopathology: 30 RA patients with direct surgical were staged by two specialized pathologists using the TNM classification system (8th edition) recommended by the American Joint Committee on Cancer (AJCC).
Data analysis: For r-MRE, real shear-wave speed (C_r-MRE) was measured on the maximum tumor level in each RA patient. For IVIM, sADC of the multiple b-value combinations was measured manually using the above methods in 20 RA patients,$$$sADC_{b1-2} (mm^2/s)=(b_2-b_1)\times\ln{(S_{b1}/S_{b2})}$$$. The optimal combination of b-values was determined by the best correlation between C_r-MRE and sADC, which decided the virtual shear-wave speed (C_v-MRE) with an equation. Then, C_v-MRE was measured on the software-generated v-MRE maps by two experienced radiologists and assessed for inter-reader agreement in 30 RA patients. Differences in the C_v-MRE were assessed between tumor (T1-2 vs T3-4), node (N- vs N+) and prognostic (P0-I vs PII-IV) stage by using the independent t-test. The diagnostic performance of C_v-MRE was compared with C_r-MRE for distinguishing T, N and P stage using the area under the curve (AUC). P < 0.05 was regarded as statistically significant.

RESULTS

The optimal combination of b-values (0 and 1500 sec/mm²）was determined with the best relation between C_r-MRE and sADC (|r| = 0.842, P = < 0.001) (Table 1, Figure 1A) and then gained $$$C_{v-MRE}=(-5.4067\pm1.1714)\times\ln{(S_0/S_{1500})}+(9.2526\pm2.3056)$$$ (Figure 1B). Figure 1C, D shows an image example from a patient. Figure 2 shows that good consistency between C_v-MRE and C_r-MRE. Inter-reader agreement for C_v-MRE was high (P <0.05). Significant differences of C_v-MRE were observed in T and P stage (all P < 0.05) exception of N stage (P > 0.05) (Table 2). The C_v-MRE for the T1-2 group were substantially lower than that of the T3-4 group. The C_v-MRE for the P0-I group were substantially lower than that of the PII-IV group. The diagnostic performance of C_v-MRE was similar to C_r-MRE in the differentiation of T (AUC: 0.813 vs 0.786; P > 0.05), N (AUC: 0.555 vs 0.545; P > 0.05) and P stage (AUC: 0.828 vs 0.799; P > 0.05) in RA patients (Figure 3).

DISCUSSION

This was a preliminary feasibility study of v-MRE for the preoperative assessment of tumor, node and prognostic stage in RA. The correlation analysis revealed the possibility of simulating v-MRE without an elastography equipment using IVIM. The results suggest that the quantitative C_v-MRE can be used to differentiate T and P stage. In T, N and P stage, the v-MRE images demonstrated comparable diagnostic performance to the r-MRE.

CONCLUSION

The optimal combination of b-values (0 and 1500 sec/mm²) can be simulated v-MRE and C_v-MRE based on IVIM has an analogously diagnostic performance with C_r-MRE, which can be used to replace r-MRE in RA patients.