Liver imaging scientists focusing on rat model of hepatocellular carcinomaHepatocellular carcinoma (HCC) is the most common primary malignant tumor of the liver with about one million deaths annually worldwide [1]. Diffuse fatty infiltration is expected to precede HCC development and may be indicative of malignant tumour [2] but there are no reliable biomarkers accurately predicting progression of fatty liver to HCCTo investigate the correlation between fat fraction (FF) measured by in/out-phase (IP/OP) imaging, with transverse relaxation time (mono- and bi-exponential T2 values) and their relationship with nodular and tumour formation in the liver of rat maintained on a choline and amino acid modified diet (CDAA diet) [3].

Six 8 week old Fisher 344 rats were placed on the CDAA diet, and 6 on normal chow. They were scanned at weeks 12, 24, 32, 40, 48, and 52 during the continuous diet using a Bruker BioSpec 94/30 MRI spectrometer (Ettlingen, Germany; 9.4T 400MHz), equipped with a gradient coil model of Bruker BGA12S, and the RF transmission and reception was performed with a volume coil ( diameter of 86mm). T2 imaging was performed using the multi-slice multi-echo (MSME) spin echo sequence, and IP/OP was obtained by fast low angle shot (FLASH). (Imaging parameters table can be seen as figure 1.

• For T2 measurements, 13 echoes was collected and spaced uniformly every 5 ms with echo times TE from 5 to 50ms, and then spaced every 20 ms from 50 to 130ms.

• Two approaches were used to evaluate and estimate FF and T2 parameters: 1) ROI-based and 2) pixel-wise methods. For nodules, a ROI-based method was used in which the MR quantitative measurements were extracted from the nodular lesions and analysed separately. For whole CDAA and control liver analysis, a pixel-wise method was used.

• T2 data were fitted to the mono-exponential decay function, STE = S0 exp(-TE/T2M) + C, [1]

where STE is the observed signal, TE is the echo time, S0 is the estimated bulk signal intensity, T2M is the mono-exponential decay constant and C and is a constant term accounting for elevation of the signal due to noise. •Data could also be fitted using the bi-exponential decay function, STE = S0 ( f exp(-TE/T2S) + (1-f) exp(-TE/T2L) ) + C, [2]

assuming contributions from two distinct T2 decay constants to the decay data. Here T2L is the spin-spin relaxation time of the T2 of the slowly decaying compartment, T2S is the T2 of the fast decaying spins, and f is the fractional contribution of the fast decaying compartment to S0.

•The fat fraction (FF), expressed as liver signal attributable to fat was determined by, 100 * (SIIP - SIOP) / (2 * SIIP) [3]

where SIIP is the signal intensity measured in the IP image and SIOP is the signal intensity derived from OP image.

1) Liver lesion appearance (Figure2)

2) T2 and FF mapping (Figure3)

3) Nodular lesion ( nodule versus surrounding liver tissue in CDAA group) (Figure 4)

Monitoring fatty changes quantitatively and qualitatively during heptocarcinogenesis in a rat model fed with continuous CDAA diet potentially provides additional diagnostic information not afforded by standard clinical MR protocols. Using a pre-clinical model of liver cancer, we have found that the mean whole liver T2L, T2M, and FF values in CDAA group are significantly higher compared to the control. T2L and FF decreased significantly with time during the diet period in the CDAA group. Further, T2L and T2M are strongly correlated with FF over time in CDAA group. For nodular analysis, the nodules were detected first at week 24; and the mean nodular T2M was significantly lower than the surrounding liver tissue at all time points. A positive correlation between T2L and FF was found in nodules. The combination of T2 values and FF is potentially a strong imaging biomarker for monitoring vulnerable tissue regions and tumor formation.