1. What are capabilities and limitations of imaging for detecting degrees of steatosis, inflammation, fibrosis, and cellular damage?
Right now you have a lot of capability in measuring steatosis. There are hundreds if not thousands of publications on proton density fat fraction. I think it's a well-validated technique, and again, limitations are going to depend on who you're using, because you may be limited in the amount of sites that you can choose based on whether the company can do both magnitude-based and complex-based proton density fat fraction.
Again, with ballooning and inflammation, I don't think there's a good way to actually measure it right now. Then for fibrosis, MRE is actually a pretty good way to measure it. MRE is also a pretty good way to measure stiffness and the limitations are more around its availability, and then just trying to get quality images out from every site in your study.
- Does the performance of the automated method vary with fat fraction? Have you considered building a method that relies only on the raw anotomical images?
The images that we use to do it are not the actual fat fraction. It's more the water and the fat image for the complex-based and for the other one it's just the six echo image. These are in-phase and out-of-phase images - what you would consider rawer, anatomical images. That's what we're using to do the segmentation. It doesn't really depend on the fat fraction as far as the accuracy. Right now we are in the midst of a study to look at a dedicated sequence for measuring liver volume versus using the proton density fat fraction sequences to compare and see what the variability is between the two different imaging techniques.
- Do you use magnitude and complex-valued PDFFs in the same trial?
Yes, that's how we are able to provide PDFF at almost every site. So before the sites started implementing it, I would say GE was probably the biggest challenge. Trying to get them to do a six echo sequence to be able to calculate the magnitude-based PDFF was a challenge because of limitations of their software. They didn't allow you to pick the echo times for your six echoes. Once they implemented IDEAL IQ, which is their complex-based PDFF method, you could get the answer out of a GE scanner. With the Siemens scanner, we can use our six echo sequence. We tend to prefer that over the complex-based, because the Siemens built-in PDFF algorithm uses the subcutaneous fat as some sort of normalization point. So in very thin patients who don't have a lot of subcutaneous fat, the algorithm fails.
The QIBA Committee on PDFF actually found this because they had a phantom that they were sending around, which was a cylinder, and in every Siemens site when they tried to do the scans, the PDFF failed because it couldn't find fat around the body habitus. That's a limitation and to avoid that limitation we just always do the six echoes at the Siemens sites and do the magnitude-based PDFF. Philips just depends on whether they've got the built-in sequence or not. The built-in sequences get essentially the same answer in the liver from both techniques.
- What would be the best way to determine cirrhosis?
Outside of a biopsy, most likely MR Elastography because it outperformed FibroScan in determining whether you are at stage four fibrosis or not. That's probably the best way to determine that.
- What are your thoughts on using MRE-based stiffness as a continuous variable in a clinical trial?
I would suggest that PDFF and MRE are continuous variables. You can't discretize them into different scales.
- Considering the correlation between FibroScan and MRE, can it be considered as the best portable system for liver imaging? In the context of trials with non-equipped sites, would it be the best option to go for?
With MRE it is stuck to the scanner, which weighs a ton, so it is not portable. But it certainly is noninvasive and MRE's probably your best noninvasive measurement for looking at fibrosis, but there are certainly some caveats. Like if the site is not equipped, then you can't do it. It's a very binary decision as to whether a site is capable of doing MRE. Either they've bought the equipment or they haven't. So if you have an MRI site that does not have the equipment, then they can't do MRE. And so then you would either have to find an alternative, like a site that's nearby or hope that you can convince the site to purchase the MRE hardware and software.
- How do you build your phantoms?
When we first started measuring phantoms in 2010, we used intralipid which is a nutritional supplement. Intralipid also contains fat, so we were able to create phantoms with it. The concern with intralipid phantoms was they contained egg as an emulsifier, and we didn't know what the shelf life would be in a longer study.
Upon researching other methods for phantoms, we decided on using t-Butanol. T-Butanol is interesting in that it's an alcohol, but it simulates fat. We can mix it with water in different concentrations. Today we are working with a company to produce phantoms that are based more in oil, which does a better job of simulating what’s actually in the liver. A benefit to these phantoms is that they will work whether we're doing magnitude-based PDFF or whether we're doing complex-based PDFF that comes off of the scanner.
- What advances need to be made in MRS, PDFF, or MRE for these to be sufficient to replace live biopsy for the diagnosis of NAFLD/NASH? For how many years could be from realizing this?
There are a couple studies going on right now. There's the LITMUS study in Europe, and there's the NIMBLE study in the U.S., which are looking at noninvasive biomarkers and comparing them to biopsy for NAFLD and NASH. The goal of these trials is to come up with a noninvasive set of biomarkers which could replace a biopsy. If those two trials are successful then I would say that it could happen, but I would say that we're probably five years away from that happening, which is unfortunate because there's sort of this wave right now of compounds that are coming into clinical trials. As you get to phase one, phase two, you could probably still use some noninvasive biomarkers and understand whether your drug is working or not, but then when you go for registration with your phase three study, you're certainly going to be using biopsies for the foreseeable future.
- Has the FDA approved imaging for NASH/NAFLD clinical trials?
There are certainly some statements on it, but it basically says that for a registration trial, you need to submit histology results. That's really how NASH is diagnosed and graded. Your results have to be on histology and the end points that they're looking for are specified and it's either an improvement in everything except fibrosis and fibrosis stays the same, or vice versa. Everything else doesn't get worse and fibrosis gets better. You currently have to use biopsies in later phase trials according to the regulatory agencies. I think that as the large collaborative studies that are out there start to produce results and people start to publish those results, then you'll start to see possibly some changes, but it's going to take a while before there's any sort of long-term evidence that the changes in imaging endpoints would correlate with improved clinical outcomes.
- Are there any additional promising imaging techniques for NASH and NAFLD on the horizon?
There are some contrast agent-based techniques that have been presented at different conferences, which are interesting because there was one at AASLD last year where they had a contrast agent which seemed to bind to fibrosis in the liver, generating an increased signal where there was fibrosis so you could get more of a direct measurement of that. There is a company out there that makes something called tilmanocept, which is a technetium-99-based compound used for bone scans. Right now they're using it for or they're trying to develop it for rheumatoid arthritis. But it actually has an affinity toward macrophages. It will be interesting to see if it actually has an affinity for macrophages if it goes into the liver and lights up where there is inflammation. There are certainly some promising techniques out there that may be off in the distant future.
Catch Up and Listen to our Webinar: