Monitoring early structural and haemodynamic changes in cirrhosis patients will provide clinicians/patients with important information on prognosis and provide a tool to assess early efficacy of novel drug therapies. There is no previous data describing the longitudinal changes of quantitative MRI parameters in liver cirrhosis. Here, we assess 1) haemodynamic and structural MR changes in stable cirrhosis at baseline in comparison to a) age/gender matched healthy volunteers and b) decompensated liver disease; 2) the annual change, over 3 years, of quantitative MRI parameters in compensated cirrhosis and its relationship to validated measures of liver injury (Fibroscan^®,¹ and Enhanced Liver Fibrosis Score (ELF)²).

50 patients were recruited to a prospective, longitudinal study of compensated cirrhosis (CC) (Child Pugh A of varying aetiology) all consenting to annual MRI visits, 10 patients with CC have completed up to Year 3 (Aetiology: 1 ALD, 2 Haemochromatosis (Haem), 3 HCV, 3 NAFLD, and 1 PSC). In addition, 10 age and gender matched Healthy Volunteers (HV) (6 male, aged 61±1 years at baseline) have completed MRI visits at baseline and Year 3, and 4 Decompensated Cirrhosis (DC) patients at baseline (1 male, aged 57±11 years). The CC group had a Fibroscan^® to assess LSM and ELF blood test every 6 months during the study. All MRI scans were performed on a 1.5T Philips Achieva system using a body transmit coil with 16-channel SENSE Torso receive coil. Data was acquired to assess structure and haemodynamics in liver/splanchnic/cardiac/renal organs. Here we show longitudinal changes related to the liver alone.

Structure: Liver tissue T₁ maps were formed using a modified respiratory-gated inversion-recovery sequence with spin-echo readout (TE 27 ms, SENSE 2, 3 sagittal slices with a 5 mm gap, FOV 288x288 mm, voxel size 3x3x8 mm³, temporal slice spacing 65 ms) at 13 inversion delays (100–1200 ms in 100 ms steps and 1500 ms). Readouts were collected at 1500 ms after the respiratory trigger synchronized to the global inversion pulse.³ Data were fit to create T₁ and M₀ maps, a ROI was drawn over the liver, and the mean T₁ determined. In addition liver volume was estimated from balanced turbo field echo (bTFE) localisers.

Haemodynamics: Vessel flow (flux and velocity) was assessed using phase contrast (PC)-MRI with 15 phases across the cardiac cycle (TR/TE 6.9/3.7 ms, FA 25^o, NEX 2, reconstructed resolution 1.17x1.17x6 mm³, TFE factor 4-6) with VENC (cm/s) = 100/50 for the hepatic artery (HA)/portal vein (PV). Liver perfusion was measured on the same 3 sagittal slices using respiratory-triggered flow alternating inversion recovery (FAIR) arterial spin labelling (ASL) with a bTFE readout (60 pairs, TE/TR 1.2/2.4 ms, SENSE 2, FA 60^o, 3x3x8 mm³ voxel, post-label delay (TI) 1100 ms, in-plane pre-saturation). Transit time was estimated from 4 pairs of tag/control images acquired at TI = 300, 500, 700, 900 ms. MR measures were correlated with LSM and ELF scores.

Figure 1 shows liver T₁ for the CC, HV and DC groups. At baseline, an increase in T₁ is found with liver disease severity (HV < CC < DC) (p<0.04), whilst a significant reduction in T₁ is observed over the 3 years in the CC group (p<0.01). Figure 2A shows T₁ measures for each CC patient and age/gender matched HV, a small standard deviation in T₁ measures across visits is seen. In comparison, Figure 2B shows LSM over visits, with a much greater standard deviation. Figure 2C shows the normalised change across visits for T₁ and LSM to highlight this further, and Figure 2D shows the poor correlation of T₁ with LSM (R=0.41). There was no significant change in PV or HA flux over visits, but PV velocity was reduced in CC patients compared to HV at baseline (p<0.01) and tends to increase over visits (P<0.006) (Figure 3A). Figure 3B shows perfusion measures for each CC patient and age/gender matched HV, in the CC group perfusion is lower than in the HV group (p<0.006). Figure 4 shows liver T₁ and PV velocity for all 4 MR visits correlated with ELF score, a significance is seen for liver T₁ (R=0.74, p<0.0002) and PV velocity (R=0.66, p<0.002).We show that at baseline, T₁ and haemodynamic parameters delineate distinct differences in liver disease severity in healthy controls, compensated cirrhosis and decompensated cirrhosis. In compensated cirrhosis, the longitudinal change of all the quantitative MRI biomarkers over 3 years shows a reduced variance in contrast to LSM using Fibroscan^®. This study provides evidence that MR may be a sensitive modality to assess changes in pathophysiology in liver cirrhosis.