that the NASH and NAFLD SIG has for you all today. I mentioned we have three great speakers and then we'll be taking questions at the end, so please put them in the Q&A box. I will note that we've already had some questions, which is fantastic. And we're gonna be focusing on the use of serum and imaging biomarkers in NAFLD and NASH today and how to use them in clinical practice. So I'm happy to introduce our first speaker, who's Dr. Emmanuel Chouhachi, who's a professor of Hepatology at UCL Institute for Liver and Disease Health, and who will be talking to us now about getting us started on this presentation, talking about using serum-based markers of NAFLD and NASH in clinical practice. Thank you very much for the introduction. And I would also like to thank ASLD for this invitation. I'm delighted to be part of this webinar. Now for the next 15 minutes, I will be talking to you about serum-based biomarkers of NAFLD and NASH. These are my disclosures, but they're not related to the content of my presentation today. And this is my talk outline. I will start with introducing some basic concepts about serum non-invasive tests, and then I will be talking about different concepts of use. This is for staging of liver fibrosis, for prognosis, for assessing treatment response, and then for risk stratification of people at risk. Now, non-invasive fibrosis markers can be broadly divided into three categories. Indirect serum tests, which are markers of liver inflammation and liver function. Direct serum tests, which are matrix components and or enzymes involved in fibrogenesis and imaging modalities of which I will not talk today. The limitations of the indirect serum tests are that they often include parameters such as bilirubin and platelets that are altered predominantly in advanced liver disease. Therefore, they are not suitable for detection of early fibrosis. And the limitations of direct serum tests are that these parameters may also reflect the presence of chronic fibrogenic disorders in other organs and systems. Therefore, they're not liver specific and we do need to interpret carefully positive results. And I will come back to this. The clinical meaning, the classic view, and this is how they were initially developed is that non-invasive tests are surrogates of liver biopsy. They are used for staging liver fibrosis. But the modern view is that they can be used as prognostic indexes to refine the prognosis of patients, but also to evaluate disease progression or even treatment response. Now, I will spend a couple of slides on the diagnostic accuracy of non-invasive tests, because although we report that diagnostic accuracy are sensitivity, specificity, area under the curve, there are no minimal diagnostic performance criteria established. So what we did a couple of years ago is that we determined the minimum diagnostic accuracy criteria with equivalence to liver biopsy in terms of mortality for cirrhosis. And we use the decision tree model to determine the minimum sensitivity and specificity of a non-invasive test to diagnose cirrhosis. The reference was equivalence to liver biopsy in terms of two year mortality. And we use the decision curve analysis to adjust the specificity of the non-invasive test because the false positive results were not adequately penalized. And what we found was that the minimum diagnostic accuracy for equivalence to liver biopsy is a sensitivity of 94% and specificity of 85%. And this means that none of the existing non-invasive tests can achieve this diagnostic performance. And we really need combination of sequential non-invasive tests. And you will see this further in my presentation. In terms of staging of fibrosis, the simple non-invasive clinical scores, they're used for detection of advanced fibrosis. And the most commonly used in NAFLD are the FIB4 score and the NAFLD fibrosis score. And you can see they consist of readily available variables that can be calculated with online calculators. And the main utility of this course is that if the value is below the low cutoff, they can exclude the presence of advanced fibrosis with a very high negative predictive value, which makes them very attractive for risk stratification in large population settings. The patented or direct serum tests, the four more commonly used are the FIBRO test, the ELF test, Fibrometer NAFLD, and ProC3. Of this, the ELF test is the most validated non-invasive test in patients with NAFLD. It consists of P3NP, TMP1, and yaluronic acid. And you can see in a recently published meta-analysis that consisted of over 4,000 patients, the ROP for F3 was above 0.8. And by using a high and a low cutoff, one can achieve sensitivities and specificities of 90%. As I mentioned before, the ELF test is non-liver specific and it can give false positive results in patients who have extra hepatic fibrotic conditions. And I think this is illustrated very well by its performance in children and adolescents, which is almost perfect because such patients do not have extra hepatic conditions and manifestations. The ProC3 is a relatively new marker. It is a marker of active fibrogenesis rather than established fibrosis. It is used in combination with other readily available variables. And you can see there are two different combinations, for instance, that all give an ROP of over 0.8. Therefore, it is promising, although further validation is required. In terms of prognosis to go to the classic view of non-invasive tests, you can see again that simple non-invasive tests, if their value is low, they can reassure doctors and patients. You can see that patients with a low FIB4 score over a 10-year period, they do not develop liver-related events. This is the same for the NAFLD fibrosis score. You can also use the FIB4 score to ascertain the risk of HCC in patients with established cirrhosis. Obviously, on its own, the FIB4 is not enough, but it can be inserted in calculators. The ELF score, it showed, this is the result from the Syntuzumab trials. And as you can see, it can predict the progression to cirrhosis in patients with F3 with a hazard ratio of four, and then liver-related events in patients with compensated cirrhosis with a hazard ratio of 2.8. So it can be used to refine prognosis. An exciting concept is the use of repeated measurements of non-invasive tests to further refine prognosis. This comes from the general population in Sweden, where they used repeated FIB4 measurements within five years. And you can see that patients that crossed thresholds from low cutoffs or intermediate cutoffs to high, or those who consistently remained at a high cutoff had the higher risk of liver-related events, or those patients where there was a change of FIB4 from baseline of at least one, had a hazard ratio of over two for liver-related events. In terms of response to treatment, unfortunately, we are not there yet. These are the results of the STELLAR 3 and 4 trials of over 1,500 patients. I want to draw your attention in the first column of histologic fibrosis responders and non-responders. Although there was a positive correlation of ELF, there are no definite cutoffs that can be drawn to predict responders or non-responders. This is a kind of a holy grail for NAFLD research at the moment, because it would simplify the conduct, the design and conduct of clinical trials. Now, for me, the most exciting prospect of serum non-invasive tests is their use in primary care or in populations at risk of NAFLD, because we can really risk stratify large cohorts and improve the early detection of significant liver disease. I will show you some key concepts before I also show you some results from a pathway that we have been working on the last five years. So, first of all, I think we all agree that the most important lesion to detect in patients with NAFLD is advanced fibrosis, because ultimately this is what has been consistently associated with liver related events. Now, this is a study from a few years back that included over 800 patients with type 2 diabetes. And you can see that they tested various non-invasive tests, FIP4, ACE2-LT ratio, APRI, ELF test and Fibroscan. And the important message from this study is that there was an over 90% negative agreement in the bottom 90% of the NAFLD subgroup, which means that these tests are fairly consistent when they give a negative result to exclude the presence of advanced fibrosis. Therefore, when we're trying to design a risk stratification pathway, these are the key concepts that one should have in mind. First of all, it should be based on evaluation of fibrosis. It should have an easily accessible first line test, which is automatically calculated, which is inexpensive and which has a high negative predictive value. And we have such a test, this is the FIP4 or the NAFLD fibrosis score. And you can see those at low risk of advanced fibrosis do not need a hepatology referral and can be rescreened in three to five years. Those with an indeterminate risk will need a Fibroscan or an ELF test. And those with a high risk need a direct referral to hepatology. Before we implemented this pathway, we modeled it extensively and based on a 5% prevalence of advanced fibrosis, we calculated a 9% refiller rate. And what we discussed with our primary care groups was that a FIP4 score would be followed by an ELF because it was more easily accessible in primary care. The initial impact was that an increased awareness and coding of NAFLD in primary care, which in itself is a huge success. But secondly, over a three year evaluation, I'm just gonna show you the results here. The detection of the patients referred 30% had advanced fibrosis compared to 8% with standard of care. So it resulted in an over fourfold increase in the early detection of advanced fibrosis. Of course, patients were not referred. We need to work on this. This is linkage to care. And it also made a perfect financial sense because you can see that the referral rate was up to 10% and the savings were enormous up to 50% per patient for healthcare system, from a healthcare system perspective compared to the standard of care. So in conclusion, there are several serum non-invasive tests available for use in different contexts. Simple tests such as FIP4 are useful for risk stratification in at-risk populations. Patented serum tests when used sequentially can reduce the need for liver biopsies and such sequential algorithms improve diagnostic accuracy. Thank you very much. And I'm looking forward to the discussion. Thank you so much. That was wonderful. And I know we'll have a lot of questions for the, at the end for the panel. So next I'm very happy to present to you all Dr. Shadab Siddiqui, who's an associate professor of medicine at Virginia Commonwealth University, who will be talking to us about image-based biomarkers in NAFLD and NASH. Thank you for the introduction and invitation. And I'm delighted to be talking to you about image biomarkers. So I've divided my talk up in four different sections. And I think before we start to actually get into discussion about biomarkers, we need to understand what a biomarker is and what is it we're measuring and what's the rigor that goes behind it so that we can translate that information into clinical practice. This will lead into discussions with FibroScan or vibration control transient elastography, MRE-based technology. And then we'll conclude by talking about what's the impact on the emerging therapy and emerging biomarkers. So let's begin by the simple question of what is a biomarker? And this is the regulatory definition of a biomarker, which is an objective patient characteristic that is measured as an indicator of a normal biologic processes, a pathogenic or abnormal biologic processes, or a biological response to a therapeutic intervention. So we have to keep this in mind as we go through this. And a biomarker must be fit for purpose. And this is important, the fit for purpose, because what that means is that when we talk about a biomarker, we have to define the context that it's being used at. What's the patient population that we're trying to use that biomarker in? What's the parameters? Is it diabetics? Is it non-diabetics? Are these elderly? Are these young patients? And then how do we test that biomarker? Is this retrospective study? Are these perspective studies? And then we sort of start to get into the nitty gritty of what are the cutoff values that are appropriate? And what are the cutoff values across the various disease entities? And then finally, we sort of talk about what's the risk of this biomarker to the patient? Because what's the risk associated with a false negative? What is the risk associated with a false positive? It's when we understand these concepts, we can sort of better define how to use a biomarker. And this is a slide that shows the various types of biomarkers that are available across various diseases and in medicine, starting all the way from susceptibility biomarker, such as genetic-based biomarkers or BRCA mutations in breast cancer, all the way up to a surrogate clinical endpoint here, which can be an alkaline phosphatase in TBC, which is associated with outcomes. But for NASH and fatty liver disease, our emphasis actually has been in clinically apparent disease and mostly in diagnostic biomarkers. So to be able to diagnose the presence of liver disease. And as we get better with therapeutic intervention and we have better therapeutic options, we may be talking about more pharmacodynamic biomarkers, being able to identify who will respond to therapy. So predictive biomarkers, looking at safety, response to therapy, but actually we're not quite there yet. So the NASH, most of my talk is actually going to focus on the diagnostic biomarkers in non-alcoholic fatty liver disease. Now, there's several things that are quantifiable on imaging biomarkers. This is liver fat, liver fibrosis, and liver inflammation. Now, each one of these topics can be a discussion for a long lecture. However, I want to focus on fibrosis. And the reason for this is this, because fibrosis is a very important predictor of survival. So to us, it matters clinically that we know what the patient's fibrosis stage is. And the study on the left shows that patients who have advanced fibrosis, meaning stage three or stage four, have worse survival than patients who have less degree of fibrosis. And the one on the right shows the more recent data from a multinational cohort where presence of even significant fibrosis, so stage two or higher, was associated with increased risk of dying or needing a liver transplant. So this underscores the central role of fibrosis as a predictor of outcomes. And this is the reason why I want to focus on fibrosis, biomarkers that focus on fatty fibrosis. So we all know vibration-controlled transient elastography, or FibroScan. Many of us use it routinely in our clinical practice. So I'm going to start by talking about FibroScan first. This is our data from the NASH CRN that included nearly 400 well-characterized patients histologically. And here we're looking at liver stiffness measurement, which is a surrogate of fibrosis spread across the varying fibrosis stages. The first thing I want to point out is that when we look at the earlier stage of fibrosis, there's significant overlap in the liver stiffness measurements. However, as we get to the more advanced stages or the higher stages of fibrosis, this tends to spread out more. And there's a greater differentiation between the mean value and median value here and the lower stages. The second thing that I want to point out is that if you look at these lesser stages of fibrosis, so stage zero, stage one, there's patients who will have really high liver stiffness measurements that are in the realm of patients who have cirrhosis of the liver, but these patients do not have cirrhosis as proven via biopsy. So these are false positive that we tend to get in patients with even no fibrosis or early fibrosis. Conversely, if you look at patients with cirrhosis, no one has a low liver stiffness measurement, meaning patients are not going to get a low liver stiffness value if they have cirrhosis or more advanced fibrosis. So again, these are not cutoffs. These are just looking at what the values are across the fibrosis spectrum. So when we look at this performance a little bit more, and this has to do with any test, we have to ask the question, what is this test being used for? So if you're focused in identifying everybody who has the disease, then we want a test that's sensitive. And so we want a highly sensitive test. And what that means is that we put cutoff values much lower. But if you're more focused on being sure about who we have, so again, this may be a clinical trial where we want to cut down on clinical, you know, false patients not meeting criteria, then we may want to be certain and we send our specificity high at 90%. And you can see for the same fibrosis stage, we have very different cutoff values. So again, how we use this test depends on the clinical question that we're asking. So these cannot just be uniformly applied to every patient. So this is where the art of medicine kind of goes in asking the right question. One thing that I want to also point out is that these tests tend to perform better as we get across, as we go higher in the fibrosis spectrum. So, you know, the AROCs tend to improve dramatically. The second thing that I want to point out with this slide is oftentimes when we use in clinical practice, and I see this happen a lot, is somebody gets a fiber scan and they say, you have stage three fibrosis. But if you look at this, the positive predictive value of these tests is actually modest at best and poor. So this is not a good test to establish the diagnosis. However, if we look at the negative predictive value, it's really good. We're sort of, you know, upwards of 90, 95%. So this is an excellent rule out test. So how do we use this in clinical practice? So if somebody has a fiber scan value, let's say of 12.1, we're 99% certain that they don't have cirrhosis. And if that's the clinical question that we're asking, then we have the answer. But if we're trying to say, does this person, can we say that this person has cirrhosis? And, you know, and if their value is over 12.1, then we're only 34% certain that they do have it and they may need a biopsy. So again, fiber scan is an excellent rule out test rather than a rule in test. This is data using meta-analysis, several studies using both mPROBE and xlPROBE. So the first thing again is there's a range in the cutoff values that have been proposed according to the patient population that is studied, the number of the spectrum of fibrosis across these studies. So the values are varying, but again, the trend is the higher the value, the more likely they have more advanced fibrosis. The second thing is, as you notice, again, across these studies, the positive predictive value is poor, meaning this is not a good rule in test, but the negative predictive values are excellent, making this an excellent rule out test. Now with any test, we have to understand the limitations of it. And for fiber scan, we actually have very well defined operating parameters. So a successful fiber scan is defined as some fiber scan where we have 10 valid measurements with at least a 60% success rate. And then the IQ or medium liver stiffness ratio is actually less than 0.3 or 30%. And this is the data from the CRN again, where we had about a thousand fiber scans done. And the unreliability, we had unreliable examined about 4% of patients and then the subsequent study, a 5% of patients. So it was relatively low, but factors that were associated with it were operator experience. So if there was an inexperienced operator doing the fiber scan, it was, you're more likely to get a less reliable result, people who had elevated ALT. So potentially inflammation contributing to unreliable result. Hispanic as an ethnicity, and this has more to do with the body phenotype and then the body mass index. So again, these are factors that kind of contribute to an unreliable result. So when we have a patient with a high BMI and an inexperienced operator, we may want to look at those results again carefully. Now transitioning to MRE or MR elastography, I wanted to show a similar slide as the last one. And this is from Dr. Lumba's group in UCSD, where we see that for lesser stages of fibrosis, the MRE values, there's significant overlap. However, as we get to higher fibrosis stage, there's less overlap and these tend to differentiate and distinguish a lot better. So again, it tends to perform better at higher fibrosis stage. And this is from the meta-analysis, including several studies with hundreds of patients. And we see a similar trend where there's overlap in the cutoff values, but importantly, the negative predictive value of MR-based technology is actually really good. The positive predictive value again is suboptimal. And this is comparing the performance of MRE versus FibroScan using biopsy as a histological gradient in a Japanese cohort. And what we see is that compared to FibroScan, MRE tends to be a better, tends to perform better, particularly in patients who have cirrhosis of the liver or advanced fibrosis. And these findings have been replicated now in additional studies in North American cohorts, particularly patients who are obese. So FibroScan performs better in obese patients, sorry, MRE performs better than FibroScan in obese patients. And this is some of the data that shows that, and this is a study that actually only included patients who were obese. And here we're looking at, the first thing I want to draw your attention to is the unsuccessful examination rate in patients who had MRE. This was about 5%. And obese patients who had FibroScan, the unsuccessful examination rate was about 20%, so significantly higher. And the key parameter that was associated with an unsuccessful examination was waist circumference with MR. But in FibroScan, there was a number of body phenotype parameters, such as chest circumference, waist circumference, BMI. And then finally, this did not reach statistical significance, but there was a trend towards skin to capsule distance. This can potentially be overcome with an Excel probe. But again, there are limitations to FibroScan, particularly in obese patients. And in those patients, maybe MR elastography would be a better modality based on the published data so far. Shear wave elastography, it uses ultrasonic B-mode imaging to assess hepatic fibrosis, sort of similar elastography, similar to other technologies. You can either have point wave, shear wave elastography, or you can have 2D where you identify an area of interest that's free of blood vessels away from the liver capsule. And in that aspect, we can actually measure elastography in these patients. And there's very limited data with 2D shear wave elastography, but this is a recent study that was just accepted for publication in CDH. And what it shows is the thresholds for detecting varying degrees of fibrosis across these three modalities. So MRE, FibroScan, and 2D shear wave elastography. And you can see the thresholds. And for detection of stage 1, 2, or 3 fibrosis, the performance of these modalities was similar. However, for detection of cirrhosis, MRE outperformed these measures. So MRE, again, tends to perform a little bit better in patients with more advanced fibrosis. This is a collation of everything that I've talked about into one slide. And what I want to point out that transient elastography is the only test which is a point of care test. So this is something that can be done in clinical practice when the patient comes to your office. And it can be done by a hepatologist, a technician, a nurse, anybody who's well trained. And, you know, through the CRN, we actually, the nursing staff was trained to do it. And they became very adept at it. So again, this is something that people can be trained easily to do. It also has the advantage of being able to measure fat at the same time as the fibrosis. So it does have that advantage. And the quality criteria of what's a good test, what's not a good test is well defined. The failure rate ranges anywhere from, you know, 3%, 4% all the way up to 20-some percent. However, in experienced hand and with the Excel probe available, that failure rate can be actually be reduced to less than 5%. And it's relatively cheap compared to other modalities. MRE is the sort of the second emerging test. And it does outperform FibroScan in patients with more advanced fibrosis, but it's not a point of care test. It's more costly. And we don't really have the quality criteria defined well for that. As far as shear wave elastography, this is something that's emerging and we don't yet have the data to sort of say one way or another how to use this in clinical practice. Now, I just want to briefly mention emerging biomarkers. And we have some data with 3D MRE where we can actually look at the whole liver for fibrosis assessment rather than just focusing on limited areas or regions of interest. We have MultiScan, which can actually link inflammation, fibrosis, and other factors to not only histology, but also potentially outcomes. And then Dr. Allen's going to talk about combining imaging modalities with serum-based approaches to actually detect varying degrees of fibrosis and disease. So these are all things that are exciting that are coming up, and that'll keep us sort of engaged. But I want to end by talking about what is priorities from not only from a clinical perspective, but just from a research perspective, what we can get. So the first and the most basic thing is that we need all of these technologies to be defined better. So what's the context of use? How do we use it? What's the parameters? And that sort of leads into the second thing is that once we get a value back on these imaging parameters or imaging biomarkers, how do we use that clinically? What does a FibroScan liver stiffness measurement of 8.6 mean to us? What's the next step? And these are things that have not been well-defined. This is an evolving field, and hopefully we'll have this better defined in the next couple of years as consortiums are set up to look at this. And then we need to be able to not only detect and diagnose the disease, but show changes in time. So what happens when a patient regresses or progresses? How can we monitor those patients? And then as we get better therapeutic options available, how do we measure the response to therapy? So FibroScan may not be the best test, or FIP4, or MRE may not be the best test in some patients, but it may be an excellent test. So we need to define which of those tests can be used in what clinical setting. And then finally, this is really important topic, which is how do we link the biomarkers to clinical outcomes? Because the way all of the clinical trials are currently designed is you have imaging biomarker. It has to be linked to a biopsy. And the reason for that is the biopsy itself is linked to a clinical outcome. So the biopsy is the surrogate. So we need data linking the liver imaging or imaging biomarkers to an outcome. So you don't have to go to the intermediary of a liver biopsy. And to show this in a pictorial perspective, this is what the current state of biomarker development is. So here, what we're looking at is patients who have fatty liver disease, who have NASH, who have NASH with fibrosis and cirrhosis. And most of us are interested in this one because this is when people have liver-related complications. So from a hepatology perspective, this is the patient population. And since patients with NASH with fibrosis are going to progress to this, we want to identify these factors. And that's where the biopsy becomes important. And this is why all the biopsy tries to link to these from an imaging perspective. But as we generate more data with imaging biomarkers, we may be actually be able to do a scan of some sort that can tell us that, you know, this patient is over here. So they're not really going to have a liver event, whereas people may be intermediate risk. And these are people who we need to target in clinical practice to improve the natural history versus we need to institute surveillance measures in these patients because they already have cirrhosis of the liver. So this is hopefully the future of biomarker imaging, which will translate into improved clinical practice and easier clinical practice. And thank you for your time. Well, thank you so much, Dr. Siddiqui. That was fantastic. And we're excited to hear more about it in the panel. And now I'm pleased to introduce our final presenter, Dr. Alina Allen, who's an assistant professor of medicine at Mayo Clinic Rochester and the director of the Nafld Clinic there. And she'll be talking about combination and sequential testing in non-invasive biomarkers. Great, thank you so much, Dr. Corey. So I will try in the last part to present the key concepts from the recent papers that combine blood biomarkers with imaging. As we heard and we know by now, there is not one ideal biomarker. There's a lot of interest now in teasing out more granular data about disease stage, prognosis, and all the other things from combining these. So that's what my talk will focus on. These are my disclosures. And to start the discussion, I'd like to bring you back to that conceptual framework of the right biomarker in the right context. When we analyze these and to anchor our knowledge about when to use combination versus just an individual serum biomarker or just imaging, I like to keep this picture in mind in terms of a disease progression type of approach. What is the goal and what is the setting in which we try to use these biomarkers? So if we were to look to the left side of this graph where the disease is early, meaning there's not a lot of fibrosis, maybe zero or one, these are the majority of our patients. And this is the context or the setting that applies mostly to primary care. So the goal in this situation is to have a biomarker that can exclude clinically significant fibrosis. So a biomarker with higher negative predictive value. As we move towards disease progression, we need to identify patients who do have some clinically significant fibrosis. So in this stage, maybe a combination of decent sensitivity with a decent specificity is what we would look for, because the goal in that setting is to identify people who need to be monitored for progression, to identify candidates for randomized control trials, and to implement aggressive management to prevent disease progression to cirrhosis. As we move to the right extreme of this figure, when the goal is to identify those with stage 3 or 4 fibrosis, this is the setting of GI and hepatology specialty. So we will use these biomarkers and think about these biomarkers in a different perspective because we want to have a good positive predictive value. We want to make sure that what we call cirrhosis is actually cirrhosis, ideally in a non-invasive way. The reason for that is calling somebody or labeling them with cirrhosis puts them on a long journey, a lifelong journey potentially, of screening for hepatocellular carcinoma every six months and making the decision to screen for esophageal varices. So we want to make sure that what we call cirrhosis is truly that. In addition, we would need to identify candidates for randomized control trials and again implement aggressive management. So keep this framework in mind to anchor some of these concepts of when to use combination biomarkers as opposed to individual ones. The other concept to anchor the combination or sequential concept is the outcome. So as it was mentioned before, a lot of the focus on development of biomarkers was on diagnosis of the disease or initial evaluation or staging the disease. And this is the main setting where the combination and sequential biomarkers have been studied as well. But other scenarios are for longitudinal monitoring. You have heard a little bit about that from the FIB4 perspective, for example, with a good example that if you follow that progressively over time, you do get some information about that. There's not a lot of imaging on that side and certainly not in combination. When we talk about prognosis or prediction of outcomes, there's not a lot of data on combination or sequential biomarkers between blood and imaging. There is emerging data on imaging biomarkers from the FibroScan perspective and also more recently from the MRE world. So there is data emerging from the combination standpoint, though not so much. So I will dive into the combination approach starting with the sequential testing. So that means an initial blood biomarker followed by an imaging biomarker. I won't delve too much into the two sequential blood biomarkers because the first talk eloquently elaborated that piece. So again, anchoring our concept on this bell-shaped curve as a framework, every blood biomarker has two cutoffs. Most of the time, we use a lower cutoff. And I'm just going to give here for an example FIB4 and nephrofibrosis score and just pointing the numbers in FIB4 since this is the most studied and validated. So we have a lower cutoff of FIB4 of less than 1.3, which when studied in patients with fatty liver disease, we can rule out clinically significant fibrosis or advanced fibrosis with a pretty decent confidence of 93% negative predictive value. We have the other extreme of a high cutoff beyond which, which is 2.67 for FIB4 and NASH. Beyond this cutoff, we have a positive predictive value of 66%, which means that in a third of the times or of the cases, what we think we call cirrhosis is actually a false positive. So not ideal, but the best we can do in terms of blood biomarkers. The problem is this intermediate zone, which in studies such as the recent one by Quentin Anstey from hepatology looking at patients from clinical trials, the indeterminate zone in this situation looking at FIB4, for example, can be in up to half of the patients. So when looking at this bell curve, we can immediately identify that on the left side of this screen, we can base our judgment easily on a simple blood biomarker such as FIB4, for example. But once we go beyond this gray zone, and then even in the red zone, we're not as confident because in half of them, we won't know what to think. And in about a third of these people, we're not sure that what we call cirrhosis is actually true. So this is the area where a combination of a blood biomarker with an imaging biomarker is prime for need and for research. And as an example, in the same recent study from Quentin Anstey, he showed that by following patients who are in this indeterminate zone with a sequential non-invasive test, it decreases this zone from about 50% to 20%. So just a brief mention of what that study showed. The sequential biomarkers that they looked were, one was FibroScan and the other one was the ELF score. And you can see the indeterminate values from the sequential approach significantly decreased to about half. And you can see the comparable sensitivity between the two approaches with the modality using imaging after FIB4 being the more sensitive one. Specificity was relatively close to each other. There is a lot of emerging data on other sequential tests. There have been really nice studies presented at the recent liver meeting. I have not seen the paper, so I did not put this in the webinar. But from the REGENERATE study showing the same approach of increasing or decreasing this indeterminate value when we follow a blood biomarker with an imaging. So I'm looking forward to that data in the upcoming future. When we talk about combination testing, we talk about pairing a blood biomarker with an imaging biomarker. And there have been a couple that have been recently published. The most known one is probably this Fibroscan AST score or FAST score published in Lancet Gastroenterology and Hepatology in 2020. The goal of this, again, thinking about the goal and the setting. The goal that this study was made is to identify patients who would be clinical trial candidates. Means that they focused on identifying those who have NASH. Not only that, but a NAS of four or higher and stage two or higher fibrosis. The inputs for this was LSM and CAP from Fibroscan along with AST. They used two cohorts. You can see the sample size here from the derivation and the validation cohort. And the concordance statistics here with 80% and between 76% and 85% for the validation cohort. I really liked this study. And I like to point out to some things that are important to keep, again, when we judge the context and the clinical use. So you can see here focusing on the rule and zone. So identifying these people. In the derivation cohort, this approach or FAST score had a sensitivity of 48%. And the validating cohort, the sensitivity of 49%. And important to keep in mind the fact that the prevalence of these outcomes, so this specific combination outcome, impacts the positive predictive value of a test. So this would be a population when we want to make sure that what we call advanced disease to enroll in a trial, we have to have a high positive predictive value. The higher prevalence of this disease, the higher positive predictive value. So you can see here in the derivation cohort, half of the patients had this outcome, a positive predictive value that was very good. In the validation cohort, the prevalence was a little lower. Therefore, the positive predictive value was a little bit lower. So again, keeping in mind the context, what's the population we studied this biomarker in? Another important point from this study was the gray zone. Again, when we have two cutoffs, a rule in and a rule out score, we have to look at that gray zone. And in this cohort, the gray zone varied between 30% to 39% between these two. Another score of combination testing is MEFIB index, which combines MRI and FIB4, again, with the goal of diagnosis clinically significant disease, F2 or higher for the role of enrolling these people in the clinical trials. And you can see, again, two cohorts, one from UCSD and one from Japan. And you can see the area under the curves from this combined biomarker being superior to these individual ones, MRE alone or FIB4 alone. Again, sensitivity was about 50% in the development cohort and 60%. So interesting, these two biomarkers will probably see more emerging data. These are fresh off the press studies in the last year or two. We cannot talk about combining two biomarkers, especially when we combine an imaging one, without talking about the cost. Because again, it matters. These can not easily be implemented at the general practice because of access and cost issues. And there's a recent paper here by Villar-Gomez, published in CGH in 2020, that looks specifically at that. So cost-effectiveness, again, for detection of cirrhosis. So looking at the most advanced of our patients. And they looked at the combinations between FIB4 and FibroScan. FIB4 and MRE was the most commonly used non-invasive test compared to FIB4 and liver biopsy, and then the individual studies here. And what they found was that the combination of FIB4 and MRE was the most accurate non-invasive and only marginally more expensive than the FIB4 plus VCTE. The reason for this marginally increasing cost is that the MRE is now orderable as a standalone test without the need to pair it with the full MRI, with the CMS reimbursement cost of $250 now. So what to do in clinical practice when we suspect a patient of fatty liver disease, specifically in the primary care level, assess the likelihood of advanced fibrosis. As we talked, I think for this cohort, a simple FIB4 is more than sufficient because of a negative predictive value of 93%. When we go above that, when we go into the gray zone and red zone, then we can't call them confidently anything. So they're either intermediate or high risk. That's where a sequential test, and what I do in my clinical practice is an imaging test of a liver stiffness measurement comes in place. There are defined cutoffs that kind of are summarized from the literature below which, like the ones depicted here, which we can call the patient a low risk. And then above which we can say either intermediate or high risk. And this is the area for us in the specialty clinic where we need to study more this combination versus sequential approach. Do we need an imaging and which one and so on? So this is where I'm very excited to see a lot more data in the future. So in summary, the setting or the goal to consider where a biomarker or a combination of biomarker fits is the first issue to consider. Are we trying to rule out disease in primary care? We need a high negative predictive value test. And in my opinion, FIT4 is good. You heard from the first talk that for people who have access to a listography, maybe we can cover more into that area too. I don't know if in the United States we completely have access to that as opposed to UK. In specialty care, we need a high positive predictive value. So in my opinion, imaging is better. You saw the area under the curves presented in the previous talk. If we focus on clinical trial enrollment, we know that serum biomarkers are inferior to maybe imaging alone and maybe a combination or sequential testing is where the field is moving. And the last thing to keep in mind is the caveat that all these combination tests were examined in tertiary centers. So the performance in primary care settings where there's lower prevalence of advanced fibrosis is unknown. And with this, I will conclude with that initial framework to say that for excluding severe disease, I would use serum biomarkers such as FIT4. For clinically significant, that gray zone, I would use a combination biomarker. And I think the field is moving towards sequential testing. And then for really defining the stage of the disease, I would use an imaging biomarker such as those listed here. And with this, I will conclude and I'll be happy to take questions. Wonderful, thank you so much. I'll invite all of our panelists to turn their cameras back on. And we have some questions from the chat. And I think one of the questions that has come up several times in the chat really is thinking about follow-up testing. And so it'd be interesting to hear from each of you, how frequently do you do follow-up testing either for both has been asked for imaging or serum-based biomarkers to monitor your patients with NAFLD or NASH? Dr. Allen. So very good question because we want to, we not only diagnose the patients, but we give them a follow-up recommendation, especially when they come to us and vice versa. Our primary care colleagues need to know, okay, so I don't need to worry about cirrhosis now. When do I need to retest? For that extreme, again, I'm thinking about that framework. What I recommend for my primary care colleagues is for patients who have risk factors for fatty liver disease, have steatosis on imaging, but don't have stigmata or features of advanced disease at this time. And they have a normal FIB4, meaning less than 1.3. I recommend to repeat that in the next one to two years if they persist to have risk factors. On the other extreme, when they come to me in my clinic and I stage them, in my practice, because of accessibility here at Mayo, I use MRE because of this, the more accuracy with higher BMI, especially for a BMI of 30 to 35 or higher, or patients with central adiposity, I go to that because it gives you a bit more granular view into, is it F2, is it F3, or is it F4? And I do like to use that now. I think we can expand it to prediction or monitoring. We had a recent paper in CGH looking specifically at that. Can we use that initial liver stiffness measurement to detect or to predict their outcomes or disease progression to cirrhosis? And what we found was that for patients who have a liver stiffness of around two, very few will progress to cirrhosis in the next five years. So what I say is repeat an MRE in five years. For a liver stiffness of around three, I repeat it in three years. And for liver stiffness of four to five, I repeat it in one year. The reason for that is each incremental kilopascal of liver stiffness is associated with a threefold increase of development of cirrhosis based on that paper. So that's the guidelines that I use in my practice based on that data on 800 plus people with MRE. Great, great. And yeah, and those values that you're quoting there are with MRE. Correct, correct. Dr. Siddiqui or Dr. Chouhatchi, anyone want to weigh in on that on how you do treat sequential testing? You know, I think Dr. Allen raises a good point. And I just want to add to a little bit. Again, this has to do with the reason for doing a subsequent test, right? So the first thing is that you want to do a test because the patient, you're worried about progression of disease. And there's a small entity of people who may be rapid progressor, but the vast majority of people, if they don't have any fibrosis, it's going to take them quite some time to get to the next fibrosis stage, especially if you start off with a very low fibrosis stage to begin with. So again, how quickly you do that is not as well defined. Usually if they have a low value, you can do it in a couple of years and you're less likely to miss them progressing rapidly. As people get higher values, then they may be more reasonable to do it sooner. But I think one of the question comes up is, what do you do in patients who are sort of in this intermediary range where they have a liver stiffness measurement on a fibroscan of let's say 6.8 or 7.2. And there's some data to suggest that if you repeat it in a couple of months or a year or six to 12 months, then you may get a better idea of which way they're heading and progressing then you can decide about what to do. So it really depends on what the initial starting value is. And if I'm worried about if they have risk factors for rapid fibrosis compression, and if they do, then I may do it sooner rather than later. But again, it has to do with the clinical question that's being asked. In a nutshell, my approach is for patients at low risk, my advice to primary care physicians is to repeat the FIB for in three years to capture disease progression or force negatives in the first instance. For the patients I see in my clinic, I don't have access to MRE and this is the case for most European centers. So I tend to repeat fibroscan and shear wave elastography annually. Great, thank you. A question came in through the Q&A and just as a reminder, please place your questions in the Q&A rather than the chat. But what about the role of liver function tests such as hepquant to predict the risk of progression to cirrhosis and to decompensating events? Is anyone using those or see those used in the future? So I can take a stab at it first. So there's a couple of these biomarkers that try to incorporate functional assessment of the liver, whether it's hepquant or it's keeping me for the time being, but there's a number of them. And these tests, what they have looked at in this, this was presented actually in abstract form a couple of years ago, where they were not able to distinguish in patients who had fibrosis NASH versus just fatty liver disease and fibrosis state. So the data with these non-functional tests exist mainly in the setting of cirrhosis or decompensated cirrhosis. And in patients who have, for instance, decompensated cirrhosis with a melt of 14, 15, 16 sort of intermediary rates, these functional tests may be a better predictor of clinical outcomes than melt alone. And this is a general hepatology paper a couple of years ago. And then similarly, the data with hepquant or functional tests, we just don't have in fatty liver disease. But I think for the most part, and most of this data has been presented in abstract form and some are sort of translated into papers that I have seen, but it can potentially identify the presence of clinically significant hypertension. But as far as differentiating fibrosis stage, as far as differentiating fatty liver disease from NASH, the data just isn't there. Great, thank you. There have been a number of questions about false positivity in FibroScan. And I know that this was touched upon, but if we could just go back and what are the reasons for false positive tests? And specifically, there was curiosity about why those who are Hispanic ethnicity may have more false positivity. And then do you interpret testing with the M-probe versus the XL-probe differently? And could that contribute? So I think that's an excellent question. And we still don't understand the reason for false positive and false negative. And I think that needs to be evaluated rigorously in the future, prospectively. But that being said, we know that there are certain parameters that go along with being able to do a FibroScan well. And I can tell you sort of anecdotal, I mean, I've shown you the data, but I can just sort of talk to you a little bit about anecdotally. It's really hard to get a good FibroScan measurement if you cannot find the intercostal space. And some of our patients in the U.S. actually, because of body habitus and small rib spaces, it's really difficult to get it into that space. Or sort of if the patient is breathing rapidly and the liver is constantly moving and the rib is going in and out of window, it becomes harder. So there are reasons for that, that it makes it more difficult. Then there's other factors that are not necessarily sort of the probe related, but maybe more sort of patient related. Everybody says that they may be fasting for X many hours before. But if somebody comes in, they may not have had the right information and they may have fasted an hour and just don't want to be, they're embarrassed and they don't want to tell you. And I say this because when we quote studies, they're very different than clinical practice. So patients may be embarrassed to talk about it and this has happened and sort of that's why I'm mentioning it. So these are all factors that potentially contribute to a false positive result, which may not necessarily be the case. And the other thing is the operator experience. So if the somebody who's doing it and this is their third FibroScan, I feel less reliable about that result than somebody who's done about 300 or 400. What we do know is that if you get a low value, irrespective of all those other parameters, you're pretty confident that that person doesn't have cirrhosis. But if they have a high value, that's really when all of these parameters and these functions start to play a bigger role in the picture. And if I could add on the Excel and then probe question, we published in 2020 on a prospective study that when with the new FibroScan machines with the automatic selection of probe, there is no different cutoffs of the two probes. So you should not interpret them differently. And I totally agree on operator experience and mils that sometimes it is underreported by the patient. We have a question about the role of L for Pro-C3. Do you all see either one of those emerging as an initial test replacing Fib4 in the future due to their higher sensitivity? Well, if you're looking at the population setting for risk stratification, it doesn't make financial sense because the Fib4 has a very high negative predictive value. So you will be adding a lot of cost with very little benefit. You can effectively rule out 60% of patients of the initial population who do not have advanced fibrosis. I don't think their role is as a first line test. And also because of the low prevalence of the target condition, you will end up with a high number of false positives. So I don't really see a role as a first line test. I agree. And I think there's good data and very interesting on the role of this score in clinical trials because it does signal something that maybe the biopsy cannot tell us. So I think that from my perspective, this is where I see as an adjunctive biomarker that will help in clinical trials to decide at one year, do we move on or do we not? Is there a signal? Maybe in combination with elastography or with other things as we move more towards the box of biopsy fibrosis stages. So that's my opinion. I think there will be a role for it, but probably not in clinical practice, specifically not at journal medicine level. Great. And I think we have time just for one more question. With all the increasing data about liver stiffness measurements, either with MRE or FibroScan, do you all see liver stiffness becoming a surrogate endpoint and replacing histology and then your future? No, not yet. I think we're still linked to biopsy because for us to be able to replace imaging biomarkers with biopsy, we have to generate data that clearly links any biomarker to a clinical outcome. And so far, we just have not generated that data. And the studies that have been done are small in number. And I think this sort of underscores how hard these studies are to do. So for instance, you would have to enroll hundreds and thousands of patients across multiple centers to be able to capture a rate of hepatic decompensation of about six to 8% per year. So that requires a lot of resources. So I just don't think we have it. But that being said, as these clinical trials that are enrolling patients with more advanced liver disease that are incorporating these biomarkers into their trial design will actually provide very useful information where we can potentially link biomarkers itself to clinical outcomes. But that data just isn't here and now. And I think designing a clinical trial only to look at that would be very difficult and would involve thousands of patients with 10 to 15 years of follow-up. Wonderful. I wanna thank our speaker so much for this wonderful webinar and thank all of our attendees for all your excellent questions and for your attention. Thank you.

serum

imaging biomarkers

non-alcoholic fatty liver disease

NAFLD

non-alcoholic steatohepatitis

NASH

liver fibrosis

disease progression

treatment response

risk stratification