THE EVALUATION OF PORTAL HYPERTENSION USING QUANTITATIVE MAGNETIC RESONANCE IMAGING (MRI)
Eleanor F Cox1, Naaventhan Palaniyappan2, Andrew Austin3, Richard O'Neill4, Greg Ramjas4, Simon Travis4, Hilary White4, Rajeev Singh3, Peter Thurley3, Indra Neil Guha2, Guruprasad Padur Aithal2, and Susan T Francis1

1SPMIC, School of Physics & Astronomy, University of Nottingham, Nottingham, United Kingdom, 2NIHR Nottingham Digestive Diseases Biomedical Research Unit, Nottingham University Hospitals NHS Trust and University of Nottingham, Nottingham, United Kingdom, 3Royal Derby Hospital, Derby, United Kingdom, 4Department of Radiology, Nottingham University Hospitals NHS Trust, Nottingham, United Kingdom

Synopsis

The hepatic venous pressure gradient (HVPG) is the only validated measure to assess portal pressure but this is invasive and not widely available. Here, we use non-invasive MR parameters as a surrogate for portal pressure. Longitudinal relaxation time (T1) measures of the liver and spleen, phase contrast measures of splanchnic and collateral flow, and ASL measures of perfusion are correlated against HVPG measures in 30 patients. Liver tissue T1 is shown to be positively correlated with HVPG (p<0.001). Combining T1 measures with splenic artery velocity using a simple linear model, it is shown that HVPG can be non-invasively assessed.

Purpose

The majority of complications in liver cirrhosis result from portal hypertension. The hepatic venous pressure gradient (HVPG)1 is the only validated measure to assess portal pressure but this requires hepatic vein catheterisation which is invasive and its availability is limited to a few specialised liver units. We aim to develop quantitative relaxometry and haemodynamic MRI methodology to estimate portal pressure.

Methods

We prospectively recruited 30 patients (9 alcoholic liver disease, 11 non-alcoholic fatty liver disease, 5 autoimmune hepatitis, 5 other, age 55 ± 13 years) undergoing HVPG measurement for clinical indications and MRI (1.5 T Philips Achieva scanner, 16-element SENSE torso receive coil) was performed within 6 weeks. Multi-slice balanced turbo field echo (bTFE) localiser images were acquired in three orthogonal directions to estimate liver and spleen volume, and identify blood vessels. The longitudinal relaxation time (T1) of the liver was measured using a modified respiratory triggered inversion-recovery sequence with spin-echo echo planar imaging (SE-EPI) readout and fat suppression at 13 inversion times (100 - 1200 ms in 100 ms steps and 1500 ms), and 3 sagittal slices acquired with minimal temporal slice spacing (65 ms) in a descend slice order in approximately 2 minutes.2 A respiratory triggered multiphase flow alternating inversion recovery (FAIR) arterial spin labelling (ASL) sequence (initial delay of 100 ms (TI) and subsequent readout spacing of 371 ms (TA)) with 6 bTFE readout phases (TR/TE 3.5/1.75 ms, FA 45°, SENSE 2, resolution 3x3x8 mm3) was acquired to quantify tissue perfusion in the liver (sagittal) and spleen (coronal-oblique).3 In addition, T1 maps of the liver and spleen were acquired with a matched bTFE readout using a modified respiratory triggered inversion-recovery sequence with 9 inversion times (100 - 900 ms in 100 ms steps) for 5 slices with minimal temporal slice spacing (144 ms) for both ascend and descend slice order, increasing the dynamic range of inversion times to (100 – 1500 ms).2 Phase-contrast (PC)-MRI was used to assess velocity, area and bulk flow in the splanchnic and collateral circulation. Velocity encoding (VENC) was 50 cm/s for portal/hepatic/azygous veins, 100 cm/s for hepatic/splenic arteries and 140 cm/s superior mesenteric artery.

Data Analysis

Each inversion recovery dataset was fit to generate M0 and T1 maps (Matlab, Mathworks) and histogram analysis was then used to assess the distribution of T1 in the liver and spleen in each subject.2 Each histogram was fit to a Gaussian function and the mode of the distribution was used to represent tissue T1 relaxation time and FWHM to assess heterogeneity. For ASL data, mean values of perfusion weighted difference signal (∆M) at each TI and TA, together with M0 and T1 were used in an iterative model to calculate tissue perfusion and arrival time.3 Q-flow software (Philips Medical Systems) was used to analyse the PC-MRI data to compute mean vessel lumen area, mean velocity, and mean flux over the cardiac cycle. Correlations between MR parameters and HVPG were computed using Pearson’s or Spearman Rho correlation coefficient (R). MR measures that significantly correlated with HVPG in the univariate analyses were assessed in a multivariate linear regression analysis.

Results

Coronal T1 maps of liver and spleen are shown in Figure 1. Liver and spleen T1 correlated significantly with HVPG (SE-EPI Liver: R=0.783, p<0.001 (Figure 2), bTFE Liver: R=0.773 p<0.001; Spleen: R=0.40, p=0.028). The strongest correlations of HVPG with splanchnic and collateral haemodynamics are shown in Figure 3. Liver tissue perfusion positively correlated with HVPG (R=0.38, p=0.046) and tissue transit time negatively correlated (R=0.467, p=0.021). Liver and spleen volume did not correlate with HVPG. The ratio of liver/spleen volume negatively correlated significantly with HVPG (Pearson R=-0.40, p=0.028). The best predictive model for HVPG included liver T1 and splenic artery velocity (p<0.001): HVPG = -26.28 + 0.038(T1liver) + 0.282(velocitysplenic_artery).

Discussion

We have demonstrated that tissue T1 and splanchnic and collateral haemodynamics correlate significantly with HVPG. A non-invasive model of liver T1 and splenic artery velocity can be used as a surrogate measure of HVPG. Various non-invasive markers of HVPG, including liver stiffness measurement, have been reported as binary predictors of clinically significant portal hypertension.4 However, the MR measures of hepatic architecture and splanchnic haemodynamics have the advantage of accurately estimating HVPG values on a continuous scale to identify the progression of portal hypertension. These findings, if validated in a larger cohort of patients, will add a valuable tool for a wider use in clinical practice.

Acknowledgements

Financial support from NIHR Nottingham Digestive Diseases Biomedical Research Unit, Nottingham University Hospitals NHS Trust and University of Nottingham.

References

1. Groszmann, R.J. and S. Wongcharatrawee, The hepatic venous pressure gradient: Anything worth doing should be done right. Hepatology, 2004; 39(2): p. 280-282.

2. Hoad, C.L., N. Palaniyappan, P. Kaye, Y. Chernova, M.W. James, C. Costigan, et al., A study of T1 relaxation time as a measure of liver fibrosis and the influence of confounding histological factors. NMR in Biomedicine, 2015; 28(6): p. 706-14.

3. Liss, P., E.F. Cox, P. Eckerbom, and S.T. Francis, Imaging of intrarenal haemodynamics and oxygen metabolism. Clinical and Experimental Pharmacology and Physiology, 2013; 40(2): p. 158-167.

4. Berzigotti, A., S. Seijo, U. Arena, J.G. Abraldes, F. Vizzutti, J.C. Garcia-Pagan, et al., Elastography, spleen size, and platelet count identify portal hypertension in patients with compensated cirrhosis. Gastroenterology, 2013; 144(1): p. 102-111 e1.

Figures

Figure1: Example coronal-oblique bTFE T1 maps for illustrating increased T1 in the liver and spleen with increasing HVPG values.

Figure 2: Liver T1 (SE-EPI) vs HVPG (P<0.001).

Figure 3: Table showing strongest correlations of HVPG with splanchnic and collateral haemodynamics.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3850