Is there a future for NAFLD/NASH trials without biopsies?

How noninvasive imaging biomarkers can replace liver biopsies

Mark W. Tengowski, D.V.M., M.S., Ph.D. – Director, Medical & Scientific Affairs at Clario

Introduction to the liver

The liver is an amazing structure in the abdominal cavity. It comprises just 2-5% of body weight yet receives ~25% of cardiac output via the hepatic artery with an additional ~60% from the portal vein. The right and left hepatic arteries deliver oxygenated blood from the aorta while the portal vein returns from the small intestine delivering deoxygenated blood and nutrients (and possible toxic substances). It is a metabolic powerhouse: transforming, storing, detoxifying and making new substances needed by the body (e.g., clotting factors, albumin etc.). The liver is composed of hepatocytes (~60% of total cell number or 80% of liver volume) with the internal vascular and bile duct system comprising the rest. These hepatocytes and bile ducts are arranged into a functional unit called the portal triad. Kupffer cells are the liver resident macrophages that are present in the liver sinusoids where they can act on bacteria or their endotoxins arriving from the gut, release cytokines, proteinoids, nitric oxide and reactive oxygen species, and finally, remove old or senescent cells. In the mouse, the arrangement of the portal tract is poorly defined histologically, comprising the hepatic artery, portal vein and bile duct, extending into the hepatic parenchyma with flow toward the terminal hepatic vein (which returns to the heart via the vena cava). This zonal arrangement of the portal lobule, acinus and hepatic lobule is discernable in human and non-human primates, and well-defined in the pig, which has lots of portal fibrous connective tissues. This zonation with its arterial and portal inputs establishes three functional zones: (1) periportal; (2) mid-zonal; and (3) centrilobular, where the functional aspects are segmented. Activities for tasks such as urea formation, glutamine synthesis, glucose metabolism and (drug) detoxification are linked with enzyme expression. OK, so why does this matter?

NASH histology

To understand non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH), you need to know what makes up the diagnostic criteria, and that means liver hematoxylin and eosin (H&E) histology from some sort of liver tissue biopsy (e.g., percutaneous or wedge resection) processed, stained and scored by a trained (hepato)pathologist. Histology is broken into three separate sections, and the Likert-style scoring based on ranges is summed, with a total score of four or more indicating a diagnosis of NASH. The first score is related to steatosis (range 0-3), and the scoring is graded on the extent of the parenchymal involvement (the hepatocytes themselves), the fat distribution by zones and the presence/absence of microvesicular patches. Inflammation is scored (range 0-3) above background levels of inflammatory cells, as assessed as lobular foci, microgranulomas, lipogranulomas in the portal areas and outright portal inflammation. Lastly, we score for the presence of ballooning cells (range 0-2). Hepatocytes are predominantly bags of smooth endoplasmic reticulum, and all that surface area is critical to its metabolic function; however, when there is cellular stress, degeneration and death, these cells balloon and increase in size (not to mention leak enzymes into the blood stream, which we can measure as increases in ALT/AST/GGT). Ballooning is scored relative to their size (often 1.5-2 times the normal diameter), frequency of few to many cells, prominent ballooning, acidophil bodies, pigmented macrophages and megamitochondria. Ballooning can be thought of an important transition observation from NAFLD to NASH and is often correlated with the increase in fibrosis. Special stains are often used to assess fibrosis — most often Masson’s trichrome and Sirius Red. Early NASH fibrosis starts with a centrilobular perivenular distribution and sinusoidal fibrosis in the perisinusoidal or periportal regions, often mild-moderate in Zone 3 and becoming portal/periportal. From here, the fibrosis worsens to be observed in the perisinusoidal and portal/periportal regions, with bridging fibrosis and eventually cirrhosis, representing diffuse fibrosis with hepatocytes confined to poorly functioning regenerative nodules. Advanced fibrosis is often considered irreversible and is frequently seen as a risk factor for progressing hepatic cancer.

It is the current thinking of the FDA that the diagnosis and staging of NAFLD/NASH can only be accomplished through liver biopsy and histologic interpretation (Draft Guidance Dec 2018 and June 2019). The FDA supports the use of noninvasive imaging biomarkers, which can replace liver biopsies in liver studies, provided the sponsor proposes diagnostic criteria that have not been established for patient selection or efficacy assessment. For non-cirrhotic and non-alcoholic fatty liver and fibrosis, even when the invasiveness and logistics of using liver biopsies in a clinical trial are discussed, the FDA recommends that only histological improvements should be used as endpoints for their likelihood to predict clinical benefit. What can noninvasive imaging tell us about these cellular events?

MRI-PDFF as a fat fraction

Intracellular triglyceride can be stained in liver frozen sections using Oil red O, but is extracted during processing to paraffin block when stained with H&E. The hematoxylin stains cell nuclei blue, and eosin (which may contain some orange G and/or phloxine B) counterstains the extracellular matrix and cytoplasm pink, with other structures taking on different shades, hues and combinations of these colors. Adipose scoring is then based on the absence of a stain, and the extent of these adipose vacuoles is judged as a function of intracellular presence and distribution over the tissue slice. The common thickness of an H&E section is 6-7 microns, and the MRI slice thickness is 5-8 mm. When you consider that a routine liver MRI has a 38×38 cm FOV, acquired with a matrix of 192×192, if the slice thickness is 6 mm, this creates a voxel ~2x2x6 mm, or ~23.5 mm³. A hepatocyte is basically a cuboid cell, 25 microns on a side. If 80% of a voxel is hepatocytes, this would represent ~1.2 million hepatocytes. Validation of MRI-PDFF across platforms and magnet manufacturers is available in the literature, and confidence is high that hepatic fat fraction can be quantified using the MRI technique. When we report the results of an MRI-PDFF analysis, our fat fraction is not accounting on the cellular level like a biopsy, but rather, on the total level of how much triglyceride is contributing to the fat signal from all cells. Using a multi-Couinaud region of interest analysis, we get a sampling/distribution of fat fraction across the entire liver, something we cannot do with biopsies for practical reasons. By using a 6 mm slice, it is possible that MRI provides a more general interpretation of the density and distribution fat present, not only in one hepatocyte but in a significant number of hepatocytes.

New 3D MRE techniques to assess inflammation

MRI-PDFF methods capture the adipose domain of the NAS-CRN score. We currently use a 2D MR elastography method to estimate liver stiffness as a surrogate for liver fibrosis. Improvements in Magnetic Resonance Elastography (MRE) techniques have identified a new multifrequency 3D elastography analysis method as a way to derive what, to this point, has been missing correlative data to biopsy histology. Current MRE methods utilize a single mechanical wave while this new technique uses 3, and it is the differences of the viscoelastic properties associated with edema, interstitial fluid pressures and viscosity that can be correlated to states of inflammation and ballooning. Research into 3D MRE has identified the damping ratio as a possible biomarker, sensitive to inflammation and possible ballooning. The damping ratio (loss modulus/storage modulus) correlated with lobular inflammation while the complex shear modulus, as a measure of shear stiffness, approximated hepatocellular ballooning. The use of 3D MRE requires further validation, but the prospect of developing a scanning protocol that derives the fat fraction from MRI-PDFF methods, the inflammation score from the 3D MRE damping ratio, the ballooning score from the shear stiffness and traditional MRE for hepatic stiffness are quite appealing as an alternative to liver biopsies.

Further reading

• Liver structure and function — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4320379/
• Liver blood supply — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5396263/
• Liver histology —  https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5760001/
• 3D MRE — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5529282/

Written by

Mark Tengowski, D.V.M., M.S., Ph.D.

Director, Medical & Scientific Affairs at Clario