Diffusion-weighted magnetic resonance imaging using a preclinical 1 T PET/MRI in healthy and tumor-bearing rats

Abstract

Background: Hybrid positron emission tomography and magnetic resonance imaging (PET/MRI) scanners are increasingly used for both clinical and preclinical imaging. Especially functional MRI sequences such as diffusion-weighted imaging (DWI) are of great interest as they provide information on a molecular level, thus, can be used as surrogate biomarkers. Due to technical restrictions, MR sequences need to be adapted for each system to perform reliable imaging. There is, to our knowledge, no suitable DWI protocol for 1 Tesla PET/MRI scanners. We aimed to establish such DWI protocol with focus on the choice of b values, suitable for longitudinal monitoring of tumor characteristics in a rat liver tumor model.

Material and methods: DWI was first performed in 18 healthy rat livers using the scanner-dependent maximum of 4 b values (0, 100, 200, 300 s/mm²). Apparent diffusion coefficients (ADC) were calculated from different b value combinations and compared to the reference measurement with four b values. T2-weighted MRI and optimized DWI with best agreement between accuracy, scanning time, and system performance stability were used to monitor orthotopic hepatocellular carcinomas (HCC) in five rats of which three underwent additional 2-deoxy-2-(¹⁸F)fluoro-D-glucose(FDG)-PET imaging. ADCs were calculated for the tumor and the surrounding liver parenchyma and verified by histopathological analysis.

Results: Compared to the reference measurements, the combination b = 0, 200, 300 s/mm² showed the highest correlation coefficient (r_s = 0.92) and agreement while reducing the acquisition time. However, measurements with less than four b values yielded significantly higher ADCs (p < 0.001). When monitoring the HCC, an expected drop of the ADC was observed over time. These findings were paralleled by FDG-PET showing both an increase in tumor size and uptake heterogeneity. Interestingly, surrounding liver parenchyma also showed a change in ADC values revealing varying levels of inflammation by immunohistochemistry.

Conclusion: We established a respiratory-gated DWI protocol for a preclinical 1 T PET/MRI scanner allowing to monitor growth-related changes in ADC values of orthotopic HCC liver tumors. By monitoring the changes in tumor ADCs over time, different cellular stages were described. However, each study needs to adapt the protocol further according to their question to generate best possible results.

Keywords: 1 Tesla; ADC; DWI; Hepatocellular carcinoma; PET/MRI; Preclinical; Rats; Tumor microenvironment.

Fig. 1

Boxplots of absolute ADC values calculated from different b value combinations: 4b = 0, 100, 200, 300 s/mm², 3b2 = 0, 200, 300 s/mm², 2b = 0, 300 s/mm², and 3b1 = 0, 100, 200 s/mm²

Fig. 2

a–c Correlation plots of measurements using 4b (b = 0, 100, 200, 300 s/mm²) plotted against a 3b2 (b = 0, 200, 300 s/mm²), b 3b1 (b = 0,100, 200 s/mm²), and c 2b (b = 0, 300 s/mm²). r_s = spearman’s rank correlation coefficient. d–f Bland-Altman-Plots showing the differences between ADC measurements plotted against their corresponding average with 95% limits of agreement (dotted lines)

Fig. 3

a Rate of HCC growth in rats. Tumor volume was measured after 10 and 20 days, as well as 13 days for tumors 1 to 4 and 16 days for tumor 5. b Representative T2-weighted images of rat 3 (tumor: white arrow) and rat 4 (tumor: white arrow) at three different time points revealing morphologic heterogeneity due to a different tumor growth rate. Viable tumor tissue is represented by hyperintense areas within the tumor, while degraded tumor tissue including necrotic areas is represented by hypointense areas. c The FDG-avid fractions of the tumor for three rats are shown. Decreasing values indicate increasing non-metabolically active tissue. d Fused FDG-PET/T2w tumor images of rat 4 (upper row) as well as threshold-based ROIs indicating viable tumor tissue for three different time points (d10, d16, d20)

Fig. 4

a ADC of the whole tumor as a function of the tumor volume. Measurements were performed after 10, 13, and 20 days for tumors 1 to 3 and, due to technical reasons, after 10 and 16 days for tumor 4. b ADC of the tumor surrounding liver parenchyma as a function of the tumor volume. Time points were the same as in Fig. 4a

Fig. 5

Hematoxylin and eosin (H&E) stained snap frozen tissue sections. a Native and tumor-bearing (rat 4) liver parenchyma. b Within tumor tissue large areas are necrotic (white arrow)