Clinical and technical evidence

A literature search was carried out for this briefing in accordance with the interim process and methods statement. This briefing includes the most relevant or best available published evidence relating to the clinical effectiveness of the technology. Further information about how the evidence for this briefing was selected is available on request by contacting mibs@nice.org.uk.

Published evidence

Four observational studies involving 428 people with liver disease are summarised in this briefing.

Table 1 summarises the clinical evidence as well as its strengths and limitations.

Overall assessment of the evidence

Four studies are included in this briefing, but further evidence was identified as abstracts (n=30). Three observational studies included in table 1 evaluate the diagnostic accuracy of multiparametric MRI using LiverMultiScan to assess people with liver disease, compared with imaging markers such as fibrosis (extracellular water) and iron content with liver histology as the reference standard. Evidence suggested that multiparametric MRI provided good diagnostic accuracy for assessing liver fibrosis, inflammation, steatosis and haemosiderosis, and could potentially improve the assessment for people with liver disease such as non-alcoholic fatty liver disease and cirrhosis.

One study (Pavlides et al. 2016) examined the use of multiparametric MRI in patients with chronic liver disease. It suggested that the liver inflammation and fibrosis (LIF) score and a direct mapping of corrected T1 (cT1) calculated using LiverMultiScan predicted clinical outcomes including all-cause mortality and liver-related events such as liver-related death, the development of hepatocellular carcinoma and new episodes of hepatic decompensation.

Overall, results are likely to be generalisable to the NHS because all studies were done in the UK.

Table 1 Summary of selected studies

Banerjee et al. (2014)

Study size, design and location

A prospective, comparative study.

UK.

Intervention and comparator(s)

MRI image with LiverMultiScan software.

Liver biopsy.

Key outcomes

Paired MRI and biopsy data were studied in 79 patients with liver disease. MRI measures correlated strongly with histological results of liver biopsy. In all patients, MRI cT1 correlated with increasing liver fibrosis (rs=0.68, p<0.0001). MRI measure of hepatic lipid content correlated with semi-quantitative steatosis scores (rs=0.89, p<0.0001). Hepatic iron content showed a strong negative correlation with T2 map of liver by MRI, indicating the diagnosis of haemosiderosis (rs=0.69, p<0.0001).

Strengths and limitations

A prospective study design. MRI operators were blinded to the indication for liver biopsy and to the patients' clinical details. The histopathologists were blinded to the MRI data. The author noted that liver biopsy is not an ideal reference standard for fibrosis, steatosis or haemosiderosis. Three study authors are on the board of directors and shareholders of the company.

Pavlides et al. (2016)

Study size, design and location

A cohort study.

UK.

Intervention and comparator(s)

MRI image with LiverMultiScan software.

No comparator.

Key outcomes

The study included a total of 117 patients with liver disease and 112 with MRI data were included in the analysis. MRI maps were acquired for the estimation of extracellular fluid and liver iron content. Liver fat content was also estimated. LIF (a direct mapping of cT1) derived from MRI was used to categorised stages of liver disease, as follows:

  • LIF less than 1, no liver disease

  • LIF 1 to 1.199, mild liver disease

  • LIF 2 to 2.99, moderate liver disease

  • LIF 3 to 4, severe liver disease.

When stratified using the LIF score, 22 patients (20%) had no liver disease, 34 (30%) had mild disease, 18 (16%) had moderate disease and 38 (34%) had severe disease. Patients were followed up after participating in the study (median=27 months), and all 22 patents with LIF less than 1 had no liver-related events found. There were 10 patients who developed liver-related events including 2 patients with moderate liver disease and 8 patients with severe liver disease. Kaplan-Meier analysis showed that patients with a LIF of 3 or more had a significantly higher cumulative risk of developing liver associated clinical complications including deaths over time compared with those with LIF less than 1 (p=0.02), and those with mild liver disease (LIF 1 to 1.99, p=0.03).

Strengths and limitations

The study included patients with liver disease, who were followed up to report any liver disease related outcomes. The length of follow-up was short to detect relevant events. Of the 9 study authors, 5 are shareholders of the company, and 2 are employees.

Pavlides et al. (2017)

Study size, design and location

A prospective pilot study.

UK.

Intervention and comparator(s)

MRI image with LiverMultiScan software.

Liver biopsy.

Key outcomes

The study included 78 patients with suspected or confirmed non-alcoholic fatty liver disease, and data from 71 patients were included in the analysis.

LIF derived from MRI was used to assess fibrosis. There was a significant association between histological fibrosis and LIF (rs=0.51, p<0.0001). A LIF cut-off of 3.0 had a sensitivity 91% and specificity 73% for the diagnosis of cirrhosis.

The LIF score was significantly associated with histological ballooning grade (rs=0.59, p<0.0001) and overall disease activity (sum of ballooning and lobular inflammation grade) (rs=0.58, p<0.0001). A LIF cut-off of 1.2 had a sensitivity of 91% and specificity of 53% for diagnosis of ballooning grade greater than 0, and a LIF cut-off of 1.6 had a sensitivity of 90% and specificity of 61% for the diagnosis of activity grade 2 or less.

Strengths and limitations

This is a prospective study. The study author noted that the use of histology as the reference standard is a limitation of the study because of sampling and observer-dependent variability of biopsy. Of the 11 study authors, 7 are shareholders of the company, including 2 who are employed by the company.

McDonald et al. (2018)

Study size, design and location

A prospective cohort study in 2 centres.

UK.

Intervention and comparator(s)

MRI image with LiverMultiScan software.

Liver biopsy.

Key outcomes

A total of 161 patients with suspected or confirmed liver disease participated in the study. All had MRI and 156 had biopsies. MRI assessment of hepatic fibrosis-inflammatory was positively associated with liver fibrosis by liver biopsy (p<0.001). There was a negative correlation between histological liver iron content and MRI T2* map of iron accumulation (rs=−0.34, p<0.001, n=142), and a significant difference in T2* between patients with and without histological iron deposition (p<0.001). A T2* cut-off of 18 milliseconds had a sensitivity of 83% and specificity of 63% for distinguishing patients with stainable iron from those without.

Strengths and limitations

This is a prospective study.

The study included 23% (n=34) patients who had liver transplant, and a subgroup analysis was conducted in the study. The use of histology as the reference standard is a limitation of the study because of sampling and observer-dependent variability of biopsy. Of the 11 study authors, 4 are employees of the company involved in the development of the technology.

Abbreviations: cT1, iron corrected T1; LIF, liver inflammation and fibrosis score; rs, Spearman's rank correlation coefficient; T2, 'true' T2; T2*, 'observed' T2.

Recent and ongoing studies