The Surfers return to diagnostics once more. This time, they explore the utility of novel blood and serum tests, assess the benefits of smaller imaging devices and consider the impact that artificial intelligence and machine learning can have in improving how we diagnose and what we learn from biopsies…one more piece of an exciting future for Fatty Liver stakeholders.
Hello. This is Stephen Harrison, and you’re listening to Surfing the NASH Tsunami .
Drug developers, investors, researchers, and corporate executives wrestle weekly to understand what is happening in commercial development of NASH medications. Join hepatology researcher and key opinion leader, Stephen Harrison, C-Suite veteran, Peter Traber, and forecasting and pricing guru, Roger Green, as they discuss the issues affecting the evolving NASH market from their own unique perspectives on this week’s edition of Surfing the NASH Tsunami.
Roger Green (00:00:35): For everyone with an interest in NASH, or more broadly, fatty liver disease, surf’s up. Episode 13 of Surfing the NASH Tsunami starts now. Hey, we’re officially teenagers. Thanks to those of you who listened to last week’s discussion of the Intercept complete response letter and what it means for the future, the most listened-to episode we’ve had so far. We didn’t get questions, we got lots of comments. I’ll chat about that in a minute. Peter Traber’s not with us this week, but Stephen and I will be joined by our usual surf buddies, Louise Campbell and Suneil Hosmane. I want to remind you, we’re still getting organized for the new patient segment, so we’re looking for questions for the webpage or email in questions to firstname.lastname@example.org. Other than that, time to get started. Let’s start with professional highlights this week. Brave one, go first.
Louise Campbell (00:01:17): I can go with a professional highlight of the week. We’ve got confirmation with somewhere around about the first of August, we’re about to start scanning in a major post-COVID study, which is a great piece of news for us, and an extensive part of adding diagnostics for liver and whether or not it could have been involved in a COVID response for the patient. Obviously, it’s a long study but it’s certainly a good piece of news for us. That’s my highlight of the week.
Roger Green (00:01:47): That’s very exciting. Suneil?
Suneil Hosmane (00:01:49): A big milestone was, internally within the company, we had a number of different ongoing R&D activities, and we finally took the time in this COVID era to sit down to go through a strategic review of a number of those activities. Now we have a lot more clarity moving on for the rest of the year in terms of some of the programs we’ll be funding and pursuing, so pretty exciting.
Stephen Harrison (00:02:13): I’m going to shift gears and do a personal highlight because coming off the Fourth of July weekend, I went against conventional protocol, if you will, and we had a mini family reunion out in the middle of Central Texas, away from COVID central here in San Antonio. Had a great time being able to get together with my parents, and siblings and their children over the Fourth of July, and watch fireworks, and celebrate our independence, which is near and dear to my heart. It was a special time, for sure. That trumps everything else I would have to say, and so that’s what I’m going to highlight this week.
Roger Green (00:02:51): For me, I think it’s probably the tremendous response to the podcast last week. We had comments come in from physicians, from industry people, from advocates. We are talking now about having co-sponsors for a regular or semi-regular patient segment. More on that in the next two or three weeks, as it starts to take formal shape. It’s exciting, and it’s a lot of momentum, and it’s really good stuff. With that done, let’s move on.
Roger Green (00:03:16): About last week, as I mentioned, no questions, several comments. Perspectives broke into two camps, kind of like they did last week, which was some people thinking that this might be the end of accelerated approval for NASH drugs in the U.S., and that would be a good thing. Others suggesting that for accelerated approval to work, you needed a clear risk-reward benefit around safety and efficacy and maybe, for whatever reason, FDA did not believe that obeticholic acid reached that level. The one consensus from everyone I spoke with was that it was a shame, as Brian said last week, how the FDA had dealt with Intercept, given how transparent and obliging and really good Intercept had been as a company throughout its dealings with the agency.
Roger Green (00:03:56): What this has done is led us to think that we want to spend the next few episodes talking about issues that will help drug developers and also drug regulators make sense more quickly of what really great agents are versus marginal agents, along the way, I think, spending some time talking about the issue of accelerated approval, but that won’t be the primary focus. Where we’re going to start is where we said we would start last week, which is unusual. Last week, I had mentioned a private investor who had asked about novel technology and thinking around diagnostics. This week, in an episode we’re going to call D Cubed, or Diagnosing Diagnostics 3, we’re going to focus on three areas where the technology of diagnostics can get better and what it will mean as it does so.
Roger Green (00:04:37): Number one, drug tests that can predict, define, or stage diseases. Number two, artificial intelligence and machine learning, and what that can do to improve our precision around what we’re looking at. Number three is in-office devices with low capital expenses, and easy to use that will measure NASH or calibrate parts of the NASH score, fibrosis, steatosis, with the kind of accuracy that we expect from larger, more expensive pieces of equipment.
Roger Green (00:05:00): One thing that complicates when we talk about outcomes, as we’ve said throughout the podcast, is that the outcome you’re looking for may depend on where the patient is in the stage of disease, so that if you’re in F3, advanced fibrosis or maybe early cirrhosis, the key outcome is stopping the progression of liver disease so the patient does not progress through cirrhosis to either long, slow, painful death or liver cancer, or a need for a transplant. A little earlier in the stage of the disease, maybe F3, certainly F2, maybe F1, the major issue is cardiovascular, because cardiovascular is a major cause of death for NASH patients, so the outcomes that you’d be looking for are drugs that have an impact directly and indirectly on how to improve cardiovascular health.
Roger Green (00:05:39): If you go all the way back into NAFLD, the issue may be around viral response, based on some of the studies that we’ve mentioned here that have to do with the relationship between NAFLD and response to the COVID-19 in a couple of settings. What we’re going to do today with that kind of triad of outcomes in mind, is each of us will address one of the three key issues I described, and we’ll ask questions throughout. Louise is going ask questions from two perspectives, as a patient advocate, and as someone who administers significant amounts of testing in FibroScan through Tawazun Health. First, let me turn it to Stephen. Stephen, why don’t you kick us off by talking about blood tests and what more we can learn from blood tests that are coming online or that we use differently over time as they are now.
Stephen Harrison (00:06:22): Thanks, Roger. This context of non-invasive diagnosis of liver disease patients with advanced fibrosis is not novel, it’s not new. It’s been around for a while. We know that the imperfect gold standard, if you will, liver biopsy, is not ideal for a lot of reasons. We’ve talked about that before, it’s well-known to everyone. We’ve got to get away from liver biopsies that are risky, that have inter- and intra-observer variability issues, into something that’s more plausible and acceptable for our patients. What we’re talking about is the context of use of identifying NASH with moderate to advanced fibrosis.
Stephen Harrison (00:07:05): The next question you have to ask, if that’s who you’re looking for is, who’s the at-risk patient? If we just apply this across the board to everybody that walks in off the street to a primary care clinic, we’ll get widely disparate results when we apply the non-invasive testing strategies that we have, compared to we’re measuring this in a high-risk population that has been well-documented to have a high prevalence of NASH. Once we know who those high-risk patients are, then we can identify the high-risk patients with moderate to advanced disease.
Stephen Harrison (00:07:39): Having said that, the rest of my little section here is going to be focused on people that we perceive to be at higher risk clinically. Ideally, that would be a diabetic, or even an obese diabetic because that’s really the low hanging fruit. That’s the population that has lots of fatty liver, and they’re enriched for non-alcoholic steatohepatitis and advanced liver disease. Historically, we’ve been able to used blood-based biomarkers, and ALT and AST have been around for decades. Alanine aminotransferase and aspartate aminotransferase, very simple chemistries that you can obtain in any basic lab. You can even do CLIA Waived testing with a Piccolo device that’s transportable and can be taken out into the community, and it gives us a very simple valuation.
Stephen Harrison (00:08:30): Where we have studied this in the context of NASH with advanced fibrosis is looking at what the AST is relative to the ALT. We call this the AST/ALT ratio. When that valuation is 0.8 to 1, that puts the patient at significantly high-risk for having advanced fibrosis. In fact, some studies suggest the odds ratio is nine for advanced liver disease, just looking at that ratio. That’s one simple test that you can do in your clinic. An example would be, you have an AST of 60, and ALT of 60 that ratio is one. That patient, by its own right is at high-risk for having moderate to advanced disease.
Stephen Harrison (00:09:12): Other newer studies have come along that have combined demographic information with biochemical information, with the idea that they’re going to improve either the sensitivity or the specificity, and then vicariously, the negative predictive value or the positive predictive value, or the overall accuracy of the test. I studied something a long time ago called the BARD score. This is built off univariate analysis for we showed that BMI over 28, AST/ALT ratio greater than or equal to .8, and the presence of diabetes all were independently predictive of that NASH patient with advanced liver disease. Then we put it in a multivariate model and applied a weighted sum where the AST/ALT ratio actually was weighted two points. BMI, one point. Diabetes, one point. If you had two or more points, you had a positive BARD score.
Stephen Harrison (00:10:07): Turns out it worked very well if you were Caucasian and lived in the Central Midwest part of the U.S., and you were obese. It carried a very high negative predictive value of excluding people with advanced disease. Simultaneously to this, the NAFLD fibrosis score was being developed by Paul Angulo at the Mayo Clinic. He found better sensitivity and similar specificity to the BARD score. Then subsequent to that, the FIB-4 test came along, which combines AST, again, with a couple of other clinical variables. Originally developed in HIV, Hep C co-infection, and then applied more across the board to non-alcoholic fatty liver.
Stephen Harrison (00:10:48): All those tests I just mentioned are very, very good for their negative predictive value, but they lack positive predictive value, unless you’re studying a very enriched population. Usually we see this in tertiary care referral centers, where people are referred with suspicion for advanced liver disease. If you do liver biopsies, the cohort of patients seen in this population is enriched for F3 and F4 disease. When you apply those four tests, the AST/ALT ratio, the BARD, the NAFLD fibrosis score and FIB-4 to a general population, you tend to get very low positive predictive values.
Stephen Harrison (00:11:27): Newer things were developed, newer blood tests were developed. You have the FibroMeter, the FibroSure, all these are proprietary algorithms that combine multiple different variables. In fact, FibroMeter version two has eight different variables, I believe, including platelet count, age, sex, hyaluronic acid, urea, AST, alpha 2 macroglobulin, et cetera. Very complicated algorithms, they’re proprietary, they’re blood tests. You draw blood, you send it off, you pay a fee, you get a result back. Turns out, they’re not much better than the simple blood test that we have, and so most of us tend not to use them because they’re quite expensive and they don’t provide a ton more utility than what I’ve already mentioned.
Stephen Harrison (00:12:14): More recently, there’s been interest looking at something that can give us a better idea about the fibrosis content inside a liver. To that end, the ELF test was developed, which combines three measures of extra cellular matrix, proteins, hyaluronic acid, procollagen-3, N-terminal peptide or P3NP, and TIMP-1. Values greater than roughly 9.8 to 10.5 were consistent with a diagnosis of advanced fibrosis. Then PRO-C3, which is kind of a cousin of procollagen-3, N-terminal peptide, but it’s a neoepitope of that, and that has also been now linked to fibrosis, particularly fibrogenesis, and values of around 15 to 16 have been identified to link a patient to advanced liver disease. Now, both of those markers of fibrosis are still under development for the context of use as a diagnostic, but those are the numbers that have been tossed around. 9.8 to 10.5 for an ELF test. It is proprietary. It does cost to send that one off and get a result back. Then PRO-C3, which is also proprietary, I don’t think is actually ready for primetime in the clinic. We use it in the research field all the time.
Stephen Harrison (00:13:38): Then more recently, and I’ll wrap up with this one. There is a new kid on the block. It is, again, a combination of variables and it’s called the NIS4 test. In this particular vein, again, we’re combining a group of four different clinical variables that can tell us about the diagnosis of NASH, particularly a NAFLD activity score of four or more with F2 to F3 fibrosis. Ideally, greater than F2, so it would include, potentially, some F4s as well. This test, like the …and I didn’t mention this before, but like the NAFLD fibrosis score and the FIB-4, is broken into two different variables, two different valuations. One with a high sensitivity, and one with a high specificity.
Stephen Harrison (00:14:33): Ultimately, that test is still needing to be validated in a couple different datasets. It was derived and validated in-house through studies that GENFIT had been doing. It compared, actually, very favorably to other tests such as the FIB-4, NAFLD fibrosis score, ELF, and Fibro-Meter. All of those that I mentioned, the NIS4 test outperformed them. There were area and receiver operator characteristic curves of .8 for NIS4, compared to 0.7 for FIB-4, 0.6 for NAFLD fibrosis score, and 0.7 for ELF, and 0.68 for Fibro-Meter. It does have good diagnostic performance, but it needs just to be validated outside of the original datasets. With that, I’ll turn it back over to you, Roger.
Roger Green (00:15:27): Stephen, what are the issues, specifically, you think we could resolve reasonably, completely using these kinds of tests, and where do you think we will not get the information that we need to be clear either on diagnosis or evaluating drugs are working, and how well?
Stephen Harrison (00:15:43): Well, we need large datasets to answer those questions. There are a couple of different areas where we can get that data. The first one is through large, independent consortia that have been set up and funded by industry, but through an agnostic channel. What I mean by that is, basically, there’s an IMI-funded initiative called LITMUS, which is mainly based in Europe. There’s a U.S.-funded consortium called NIMBLE, that has NIH funding attached to it. Both of these entities were established to identify non-invasive testing strategies. Not just blood-based or wet biomarkers, but also imaging biomarkers, and then potentially the combination of wet biomarkers plus imaging.
Stephen Harrison (00:16:35): Those two entities were set up to study, literally, several hundred to thousands of patients, to analyze in a very agnostic manner, these non-invasive testing platforms that I’ve mentioned, particularly NAFLD fibrosis score, FIB-4, ELF test, PRO-C3 and now NIS4 is certainly being analyzed, I believe, in LITMUS and I think, I’m not sure about NIMBLE. That’s one way that you could get at this answer as far as a validation. The other one is focusing on large, phase three trials, or very large phase two B trials, where pharmaceutical industry is actually measuring these non-invasive tests as either secondary or exploratory endpoints as part of their large pair liver biopsy clinical trials. While it’s not a primary endpoint, this data is being analyzed alongside the biopsy for its correlation coefficient, for its accuracy as it’s able to predict the presence of NASH with advanced disease.
Stephen Harrison (00:17:46): We’ve generated some data. We’ve seen large datasets from Gilead come out giving us the data with ELF that I’ve mentioned, where we found the score of 9.8 predicting advanced liver disease. They’ve also shown that a score greater than 11.3, different context of use, but is associated with progression to decompensating liver disease. These pharmaceutical large datasets are very helpful and instructive. Ideally, we need more of them, and we need other companies to validate what one company shows. At the end of the day, it’s all about numbers and generating enough data to substantiate the claim that’s been made.
Roger Green (00:18:28): Louise?
Louise Campbell (00:18:29): I think from a perspective of how patients view this, or how we can certainly see how the timeline’s gone. I think Stephen’s absolutely correct. AST and ALT has been around for a long time, and the abnormalities in that. If we look back, in 1950, cirrhosis was noted in diabetics. We are now 70 years on in that timeline, and we are still discussing the poor diagnosis of patients with fatty liver disease. The predominant ones that Stephen repeatedly says are the low hanging fruit, which is diabetes. If I go and look at the British Liver Trust’s survey from last year, it was one of the largest surveys ever carried out in the UK, only 4% of patients being diagnosed in a GP surgery are those who are being picked up with lifestyle conditions, so that’s not diabetes, that’s not cardiovascular disease.
Louise Campbell (00:19:29): In those 70 years since we detected that cirrhosis was noted in diabetes, and despite the mortality rate of diabetics increasing due to liver disease, or renal failure, and hypertension, and all of the other comorbids, we’ve not got any better at it. All of these blood tests have been around and available, but we are left with exactly the same discussion from several weeks ago, we have, and we’re desperately trying to find blood markers in an organ that doesn’t show distress. When we’re looking for conditions in liquid biopsies, we’re looking for the condition itself, so we’re looking for a tumor.
Louise Campbell (00:20:18): Liquid biopsies, obviously, are very well tried in hepatocellular carcinoma, so we’re actually looking for the other condition to give us a symbol and a sign so that we can desperately find it, despite having all of these tests available and despite having great outcomes for some of these tests. FIB-4, if used in primary care very well, can actually reduce significant referrals to secondary care, but actually, we’re still looking for abnormal tests.
Louise Campbell (00:20:47): As we know, about 70% of patients with abnormal livers will have normal liver biochemistry, so it is about getting better, but it is about asking the question in the first place. Because an awful lot of people have gone through medical, nursing, nutrition, dietetics training in that 70 years, but I regularly hear from GPs, when I look at Twitter, when I look at these nutritional forums, that they do not get nutritional input into their training. They’re not looking for a disease, as we know, affects one in four of the global population.
Louise Campbell (00:21:28): Stephen’s right. A lot of these blood biomarkers only target the patients whose ethnicity is Caucasian. If you look at Asia, of course will have a different definition for NAFLD, and NASH on their BMI, their BMI is less. We know that Hispanics have a different rate, so there’s a lot of ethnicity profiling that needs to be done so that all of these testing systems actually take into account the genetics of patients.
Louise Campbell (00:21:57): When I was treating Hepatitis C, our Arabic population reacted very differently and Sub-Saharan population reacted very differently to Interferon to a lot of Caucasians, so you had to take that into account when you were treating them. Everybody is an individual. Patients want to be treated as individuals, but biomarkers, yes, but we’ve got to be more specific. We’ve got to say that you don’t… and EASL guidelines, AASLD guidelines, I think, very clear that you do not use liver function tests to discount liver disease of any description. You must go further than that. If we’re only using blood biomarkers, then we’re missing a vast population, as we know.
Stephen Harrison (00:22:40): Let me just expand on what I was saying earlier, that might take this a little further than where we’ve been in the past. If you were just to apply something simple like FIB-4, and Louise, your point about reducing the rate of referrals using that is an accurate reflection of what’s happening in the UK, having talked to several of my colleagues there, and to some degree, it’s happening in the U.S. as well. One area where you could use that, potentially, effectively is let’s say you’re an endocrinology clinic and you’re seeing a bunch of diabetics. Well, that’s a high-risk population, so you could take a score… It doesn’t have to be the FIB-4. It could be the NAFLD fibrosis score, it could be the NIS4.
Stephen Harrison (00:23:27): By the way, I didn’t mention what’s in NIS4. It’s alpha 2 macroglobulin, a microRNA called miR-34, HbA1c, so a measure of glucose levels, so hemoglobin A1c, and the YKL-40. Those four markers make up NIS4. You could take any one of these that have a low cutoff and a high cutoff, particularly the ones I mentioned, so NIS-4, and NFLD fibrosis score, and FIB-4. If the patient that is high-risk, such as a diabetic, is lower than the low cutoff, then you can feel confident that that patient, while they have risk factors, is certainly not at risk for liver disease right now, and can be deferred in their evaluation.
Stephen Harrison (00:24:12): If they’re above that low cutoff, then that patient needs something else done, either a sequential test with what we’ll talk about in a moment, an imaging study, or maybe a second blood-based biomarker test, or you could just refer that patient to a hepatologist and let them move forward with their own evaluation. Then, obviously, if you’re at the high range that’s easy, because those patients get referred on anyway. It makes it very simple. If you’re dealing with a high-risk population, so maybe let’s just tell … If you’re listening and you’re an endocrinologist, you’re a primary care doctor, you can apply those three tests very simply to a diabetic population.
Stephen Harrison (00:24:54): If you’re lower than the lower limit, for FIB-4, it’s 1.3, then you’re good. If you’re higher than that limit, they need another test. Either another blood test, or an imaging study, or a referral to a hepatologist. I think that gets us beyond where we were 50 years ago. At least it gives us another benchmark to achieve. It’s not certainly the end-all, be-all. It does identify patients that would otherwise not be diagnosed because either their liver enzymes aren’t hitting a red flag, or they’re not complaining of a symptom, or the primary care doc just doesn’t have time to focus in on fatty liver today. They did the simple test, simple blood test, they could get that answer and then refer them on if it’s abnormal.
Roger Green (00:25:39): Suneil, noting what you do for a living is you oversee the commercialization of NIS4 for GENFIT, is there anything else to say about how they’re being used beyond what Stephen and Louise have been talking about, in terms of the value they can bring?
Suneil Hosmane (00:25:54): Many of the important points have been discussed, but I’ll try to bring them together and try to integrate into, for example, my thinking of the space. First and foremost, fully appreciate the point from a patient perspective that time has gone on, yet we’re still not completely there. I think that one thing that will greatly help … There’s many factors that influence that. There’s economic factors that influence that. There’s obviously past performance characteristics that influence that, and sophistication of guidelines that influence that.
Suneil Hosmane (00:26:25): If we just take that last one as a major catalyst that we believe is going to be influencing how we approach this moving forward, if you really dig into and we take the NICE guidelines aside for a second, they’re very unique in that regard, and you say, “What do AASLD and EASL guidelines look like?” They provide context, but they don’t go that last step in providing the granular details that are necessary to implement a screening strategy or use of test one, two, three, four to finally get at that evaluation. I think it’s very important, so if a guideline says, “Use FIB-4, use x, y, z to identify patients,” but doesn’t really provide the guidance as to how, which cutoff to use and which sequence, and more importantly, what data supports that algorithm … I think that’s the thing that Stephen alluded to, which is all about the validation, and bringing in bigger datasets to bear and inform on what’s that right strategy.
Suneil Hosmane (00:27:30): I think that’s the missing piece that’s going to really, I think, move the needle. Because when it comes to taking the good preliminary data and then moving that one step forward into the clinic, it’s all about clinical utility. In order to really define that, we’re talking about, how does that testing paradigm influence clinical behavior? I think, for example, as someone in that field, we’re committed to doing that. We’re committed to funding that. It’s an endeavor. It’s an endeavor and it takes a while to generate that evidence, but once you have that evidence, you can look at a health system. You can show this kind of flowchart of how patients move in the system as a result of various testing strategies, and what does that mean? What does it mean for them? What does it mean for the healthcare system? What does it mean to the overall cost that’s borne on those patients? I think that’s very important.
Suneil Hosmane (00:28:24): I’ll just make a very small comment. NIS4 is available through LabCorp and Covance, but that’s just for clinical trials and research. We haven’t finalized a deal to move into the clinic just yet, but we’re working very, very hard on that and we hope to have an announcement on that fairly soon, but just wanted to caveat that.
Roger Green (00:28:44): Thanks, everybody. We’ve talked about liquid biopsy. The other meaningful source of data, without biopsy, would be the kinds of imaging devices that are available. Historically, those tend to be things that are either very expensive and are a little bit cumbersome or bulky to manage, FibroScan, and costly. One of the areas that we’re seeing development is in the creation of lower cost, high quality imaging devices of different kinds. Suneil, as the second piece of where data’s likely to be generated in new ways, I’d like you to spend a little bit of time, if you could, just talking about some of those. What’s coming up and where it’s coming from.
Suneil Hosmane (00:29:22): No problem. I think it’s a very exciting time, and I think there’s a lot of things fueling this. If we think about the revolution of what’s happening from a mobile computing standpoint, cellphones, essentially, far more powerful than super computers of a few decades ago, and of course, at a very, very low power consumption. This really enables really, really interesting things. I think 5G, I think this is going to be taken to a new level. I would say, just taking a step back to put it into reference, an MRI … By the way, I’m not vilifying MRI, I think it’s a phenomenal tool and it’s done wonders for medicine.
Suneil Hosmane (00:30:01): Generally speaking, whenever you say, “One tesla, two tesla, three tesla MRI systems,” each tesla costs about a million dollars, so just to put it into perspective. It’s unfair to suggest that a smaller system or point of care system exactly matches and outperforms an MRI. I don’t think it’s intended to do that. However, if it can get closer, either through innovations on a hardware standpoint or innovations on a software standpoint, that’s tremendous value to the healthcare system, to the patient, who now may not need to book and wait for a time on the scanner, which is certainly an issue that we’re hearing about with COVID, for example, and there’s delays to getting access to some of these systems. Nevertheless, this democratizes the technology and provides better insights to triage, to identify and bring some of that power back to the clinician, the healthcare provider at the clinic, ultimately empowering the patient.
Suneil Hosmane (00:31:01): If I could describe some of these developments, we can put them two axes, some that are more on the software side of innovation and some that are more on the hardware side of innovation. I’ll start with hardware. As you know, there’s been a movement towards miniaturization. You go in point of care, and ultrasound has slowly been doing this over the last several years, decades in fact. The classic trade-off has always been around quality. As you go to a point of care system that runs off a battery, it’s a low-power system, can you get the same fidelity as a large system? Generally, the answer in the past has been, “No,” but with some of these innovations that are occurring by smaller companies …
Suneil Hosmane (00:31:41): I’ll give you a few examples. SonicIncytes, Esonic is another example of point of care Butterfly network. These types of systems that miniaturize the system, but yet keeping a very high fidelity scan. Now you’re not really making as much of a compromise as going to a full-out radiology system. Through software, they’re able to do all kinds of unique implementations to extract features and information about tissue, tissue function. In some cases, even looking at heat. For example, caloric output. There’s a company called ENDRA that’s doing this kind of work on this axis, more from a software perspective.
Suneil Hosmane (00:32:32): I think what that’s doing is it’s taking some of those features that normally would have been in a large MRI type facility, and democratizing them down into a point of care system, and taking into account ease of use. I say that because that is a linchpin for this field, where you can make things smaller but you may not make them easy to use, and so there’s a lot of smart software that has to go in on the backend to ensure that you’re getting the right measurement, you’re looking at the right organ, and that this technology’s easily scalable, trainable, and things of that nature.
Roger Green (00:33:15): Stephen?
Stephen Harrison (00:33:16): Where we’re headed is definitely in that direction. We need to be able to go to the patient. Make it easy for the patient to get tested, rather than them having to seek out, get in their car, look up a place to go, wait in a long line, find a time that’s convenient for them that matches the convenience of the imaging center. That’s the way we do it today, or that’s the way we’ve historically done it. Just think back to you have belly pain, you go see your doctor. He orders an MRI, or a CT scan, or a right upper quadrant ultrasound. You leave the doctor, you go to the scheduler, they book you an appointment. Then you have to go. Sometimes, you have to take a prep.
Stephen Harrison (00:34:06): Sometimes, you have to have an IV put in, and you get contrast, plus or minus. Then you get the study, and then you have to wait a while for the results, sometimes up to a week for the radiologist to read it, send the report to the primary care or the referring doctor that ordered it. Then they have to call you and go over the results. Often, that’s done by a physician extender, not even the doctor necessarily that ordered the test. Could have been a medical assistant that’s helping the physician as well. If you have questions about it, sometimes you have to leave those questions, and then a doc finally gets back to you, or you wait until you see the doctor in clinic to go over the results.
Stephen Harrison (00:34:48): Having said all that, we need to get better at that if we’re going to address Louise’s concern of finding these people with their asymptomatic disease, but have high risk or have already developed advanced liver disease and just don’t know it. That brings in these point of care tests, if you will. These hand-held devices, these minimal footprint devices that can go into the clinic, or that can sit in the corner of a clinic and be pulled out and a result can be provided within a matter of minutes to the patient. It’s non-invasive, it’s easy to use, it’s easy to interpret, and you get a result.
Stephen Harrison (00:35:30): Now, that result doesn’t necessarily have to be as detailed or quantifiable as a fancy MRI test, but it needs to give us a ballpark. Does the patient have fatty liver? I don’t need to know how much fatty liver. Does the patient have fatty liver? Does the patient have at-risk amounts of fibrosis. I don’t need to know if it’s an F2, if it’s an F3, or if it’s really even an F4. I just need to know that that patient has some degree of scar tissue that warrants further evaluation, because they are now deemed moderate or advanced. Meaning, essentially, F2 or greater. I don’t even need to know if there’s underlying NASH present. I’m happy knowing there’s fat and there’s some degree of fibrosis. Because you don’t get fibrosis without inflammation. That’s all part of the wound healing response to injury.
Stephen Harrison (00:36:25): If a point of care test can provide that for me, I think that’s great. I think that takes us to the next level, the next step, if you will. Now we can provide a simple blood-based test. We can marry it with a simple point of care imaging test. We can get pretty darn close to where we need to be on who should be referred for further evaluation, and who should be just told, “Lose weight and exercise. Come back and see me in six months.”
Roger Green (00:36:54): Fundamentally, what you’re describing is a better system, basically, to weed out of the system people who don’t need additional testing or care, so that we can do a better job of using time and resources to focus on folks who do need more care. Is that where all that comes together?
Stephen Harrison (00:37:11): Yeah. Again, if you just take for simple math, 100 million Americans have fatty liver and really, we think by most accounts, only about 25 million of those, that’s a huge number, so I don’t want to downplay that, but 25 million of the 100 million are at risk of disease progression, then if we have a test that can exclude the 75 million that we don’t need to worry about today, doesn’t mean we won’t ever have to worry about them. We can talk about that another day about when we need to do follow-up testing.
Stephen Harrison (00:37:45): Today, when they’re in my clinic, if I have a test that can exclude 75, 80 million of them, that’s the 75, 80% solution right there. Then we can hone in on the those that have a positive test, and that’s where we still need better diagnostics, because getting at the people that are at risk and defining with some degree of fidelity how severe their disease is without sticking a needle in their liver.
Roger Green (00:38:15): Which we’ll get to in a minute. Louise, how does that line up for you with what you do right now with Tawazun and FibroScan? How would this make your work easier or just enable more people to get care in ways that are helpful to them?
Louise Campbell (00:38:28): I think Stephen makes an excellent point there, and it’s one of the reasons that Tawazun Health was created. As you know, medical healthcare is all about the few, the actual few that got the test that was abnormal. The actual few that were picked by the GPs to send to secondary care. I think by targeting with non-invasive technologies, the new devices coming as discussed earlier like FibroScan, and I use FibroScan predominately because it was the one that was most researched. It’s the one that’s biopsy proven. It’s the one that’s available now. As we know, in the next couple of months, it’s only going to get more accurate with release of new systems, which will bring in more line with high quantification, high quality for the CAP, high quality for the kilopascal, so it’s only getting better.
Louise Campbell (00:39:19): Stephen makes one absolutely great point in the context of, why should this be healthcare? A lot of people go to the gym and make significant changes to their lifestyle because they know their lifestyle problem. If I was a dietician today, I would not be doing weight loss or dietetics without a FibroScan or an equivalent device. Hand-held, able to be used. If you’re going to advise somebody on diet, weight and to plan that management, we know that you’d want to be able to prove and watch that change in behavior.
Louise Campbell (00:39:52): Stephen said a couple of weeks ago that the average engagement is about three months. One of the advantages of these new devices and the smaller technology is they’re highly repeatable. If we take FibroScan, people say it’s an expensive test. Well, so is a lot of tests when we get to hospital because, certainly in the UK, we go through a fee paying system at each stage. Every GP assessment has a cost. Every specialist referral has a cost. A gastroenterologist is slightly cheaper than a hepatologist, but to actually get access to FibroScan costs the NHS and every CCG, who are the clinical care commissioning groups for the GPs, around about 500- 700 pounds per patient, so there is no surprise that this technology is held upstream.
Louise Campbell (00:40:47): You’ve got to pick up a patient by a blood test, probably FIB-4, ELF, or one of the ones we discussed to get access to a FibroScan. Actually, if you FibroScan or use these non-invasive technologies earlier in the time point, you won’t miss the person who is going to have the normal FIB-4, or ELF, or AST ratio. What you’re using is a funnel system. The more people who get access to it further down the stream, the cheaper the system becomes. Therefore, it becomes more cost effective to be able to do that, which is one reason that Tawazun Health was created. I can go around, and our operatives can go around, any GP surgery at minimal cost to that pathway, and it doesn’t matter what device is being used.
Louise Campbell (00:41:38): In some respects, it’s about patient engagement. Blood tests and liquid tests do not engage patients. Being able to sit there with the patient in front of you, and that’s one of the reasons that I’m engaged very heavily with FibroScan, because in 35 years of dealing with liver patients, or alcohol dependency, or any drug dependency, the actual engagement you get from the patient, whether it is a child, to the mother, to the father, their engagement and their change is quite dramatic. If you follow up every month or every three months, particularly if you’ve got them on a timeline, you keep that engagement, and you can say, “Did you change something, or did you not change something?”
Louise Campbell (00:42:22): I think, I would never give a patient a result back that I’m following up in FibroScan until I say, “What have you been doing for the last three months?” It is very, very rare that somebody will tell me that, “I made a change and I don’t see an improvement.” The other option is the nine out of 10 who do change aren’t the problem. It’s the one or two out of 10 who don’t change, that if you catch them at that three months and say, “Yeah, I can see you didn’t make that change,” then that’s really important because then, it’s like, “Oops.” They reconfigure and they re-think about that whole process.
Louise Campbell (00:42:57): I’ve got a great example. I had a gentleman who changed jobs, and he started to work into a hotel. I did his FibroScan, and literally, it was, “Oh my god. What have you done?” He said he’d now started eating fast food every single day because it was easier, but his liver had gone to, well, very fatty from relatively normal, to a high stiffness score, which obviously, doesn’t necessarily equate with fibrosis, but the liver function test highly indicated that he had a highly inflamed liver, and you can see it quite rapidly.
Louise Campbell (00:43:37): There is a lot of doubt out there as to whether or not serial readings with FibroScan can be predictive or not predictive, but I think that’s down to the user. I think that’s down to the quality of the scan and the IQR. That’s going to be the same in any device. The more devices that we can get at a local level, the cheaper the whole system becomes, and the less we’re going to miss people. Because liver cancer is the second largest cancer in the world. It is diagnosed so late, that only one in 10 patients survives more than five years, despite all of the discussion we’re having today about these techniques, it is still for the few. The few that have actually been diagnosed, not the ones that we need to get out there.
Louise Campbell (00:44:24): If I was a gym, if I was Jenny Craig, if I was Weight Watchers over here, if I am putting in a weight loss program, I would be wanting to know and add to my system something that can show those people that they can still carry on losing weight, get fit, and actually reverse liver health. Because you look at heart diets, it’s not about the heart. The heart’s a muscle. It’s about improving your internal metabolism, your liver health. Diabetes diets. Again, it’s not about the diabetes, it’s about the improvement of the liver health, but we’re not testing for it. Everybody’s targeting diet and improvement. Nobody’s measuring it, except in blood markers. I think what we miss then is that ability to keep a patient on track. If we can keep the ones who can be kept on track and really get that motivation, then we’re left with the ones that are harder to treat, the real behavioral changes. The ones that are too far down that pathway, that we might need bariatric surgery for.
Louise Campbell (00:45:28): I think, I wasn’t part of the discussion last week, but I thought it was a really engaging session. I think if we look at what the FDA was asking for, which were clinical endpoints, if we look back with fat in the liver, we know cardiovascular risk increases with simple steatosis, but we’re not looking for it because we’re targeting drugs and F3 and F4, just before cancer. It is a sad shame that patients weren’t involved in the discussion with the FDA, because people will now be diagnosed with cancer because we can’t give them a drug that could reduce or slow that process down while we wait for this.
Louise Campbell (00:46:15): We’ve seen it in Hepatitis C. We saw a great turnaround in mortality, transplant costs, donors being needed because we treated the sickest Hepatitis patients first with the orals, so we have the evidence that if we can actually change the end pathways for liver disease, we can still have great outcomes. While Intercept and other drug companies now accumulate that data of how many less days those patients spend in hospital, how many times they less visited the GP, how much more time they spent at work for these endpoints, people will die, or will develop liver cancer.
Louise Campbell (00:46:54): I think, agreeing completely with Stephen, we need a combined effort throughout the country, and all countries, that these devices shouldn’t necessarily be only medical, because they are going to be so useful to every gym, particularly if you’re on steroids. You’ve got a little inflamed liver, but you look really healthy on the outside, but actually, what are you doing internally? That’s why I engage with FibroScan. That’s why we developed Tawazun Health, to make this cheaper to doctors, to primary care, to hospitals, to allow all sorts of patients into clinical trials, for example.
Louise Campbell (00:47:36): I heard a very good webinar by Perspectum last week, and a lot was said about why FibroScan shouldn’t be used in comparison to Multiscan. FibroScan is not a direct competitor to Multiscan. If you’re going to find the patients you need to put through Multiscan, FibroScan is going to find the mass. Multiscan’s not going to do it, so there’s a combination of all of these devices and wet markers, and AI, which we’re going to get on to, to help patients, but find more, but it doesn’t just have to be in a medical world. We don’t isolate genes for that.
Roger Green (00:48:17): First of all, duly noted. Louise, since the first time you’ve came on, you’ve been asking, the question isn’t, “Who should we test?” Its, “Who should we not test?” Implication really, we should test everybody or virtually everybody. Are we yet at the point where we have digital biomarkers that come out of equipment, or will come out of the next generation of, I guess, you mentioned ENDRO, SonicInsights, Esonic, the others. Will be able to create some kind of digital markers from there that people can use to gauge their health, or are those devices not precise enough to do that?
Suneil Hosmane (00:48:43): I think so. I think, in fact, as I talk to some of the founders and some of the people in that space, they refer to that as physical biomarkers, so it is yet a biomarker. There’s many different combinations that you can create, you can find, you can identify, leveraging the next optic as well, which is AI, machine learning vision as well to extract those features. I think the answer is, “Yes.” It will take a little bit of time to develop the evidence for that.
Suneil Hosmane (00:49:13): To all the points that were just discussed, I think the big benefit of some of these new technologies is the cost. I think that is an elephant in the room. If you have $100,000 piece of equipment, you’re fundamentally limited to where you can put it, unless there’s someone paying for that or there’s a direct cost payback period, I think it’s very prohibitive for a lot of clinicians, clinics, and even professional groups, including dieticians, and stuff like that. At least from some of the discussions that I’ve had, at least behind closed doors. The interest is there. I think, as some of these innovations come out, the cost comes down, more of the things are enabled, I think it can lead to that world that we just described.
Roger Green (00:50:01): Thanks, Suneil. Let me follow up with one thought, and then I want to go on to AI, because we’re going to run out of time quickly. What Louise is describing is what, I guess, I describe sometimes on the podcast as the wellness paradigm, which is helping people be well. When you talk to physicians sometimes, they say, “Well, I only test once a year. Why would I do more than that?” Because the liver doesn’t recover all that quickly, and drugs don’t work all that fast, but if the purpose is to help the patient guide her or his own health, then things that we can test more often to give people data to say, “You’re on the right track.” “You’re on the wrong track.” “What track do you want to be on?” Then, work with them on how to get there.
Roger Green (00:50:34): Stephen talked about this final phenomenon, that people do their best for three months and they start to come back up. What I think we’ve learned from the wellness sources is that if we work on behavior and we provide data more frequently, then we have a better chance, far more cost effectively than in the healthcare system of helping patients figure out ways that they can drive and improve their own health using steps that they’re willing to take, rather than those that are imposed on them. Just an editorial comment. Any thoughts on that?
Suneil Hosmane (00:51:00): I’ll jump in just really quick. I think that’s spot-on. I think about it in a larger context of gamification and the gamification of healthcare in many ways. I think about just my Apple Watch, not that I’m promoting the Apple Watch. I get zero incentive from Apple. Just as a discussion topic, if you have one of those, you see the rings and you know intuitively if that day you were healthy, you’ve been mobile, you’ve been doing stuff. You know if you’re active or not. It’s very different, at least to me, when I see the rings and they’re not even remotely closed, and I know I’ve been super lazy that day. There’s an incentive to get on the bike, do something. I got three hours left in the day, “What can I get done?” Going up the stairs a little bit, anything.
Suneil Hosmane (00:51:46): I think that’s kind of where we’re talking about in terms of patient engagement and these sort of, whatever you want to call it, blood-based, imaging-based, whatever the case may be. These data points and these quantifications that can provide additional guidance and evidence to help us lead the better version of our own lives. I think that’s super important.
Roger Green (00:52:07): You say Apple, I say Samsung, same idea. You’re absolutely right. It’s not the product, it’s the general approach to the idea of wearables that I think matters more. Stephen, do you have anything to add before we jump into AI?
Stephen Harrison (00:52:16): I think those points are all very valid. I agree with them 100%. We know in clinical trials, the more frequent visits we make with our patients, the higher the probability that they’re going to do what we ask them to do. Essentially, whatever device you have, a Fitbit, an Apple Watch, whatever, becomes that reminder. It absolutely is true. It just makes me feel bad though, every time I look at mine and I don’t do what it tells me I’m supposed to do, and I don’t have the time to go do it, so I just feel bad about it. It is helpful to some people.
Roger Green (00:52:49): I know what your workouts look like five days a week. I’m not really worried about you, personally.
Louise Campbell (00:52:52): I do have one comment come into that with Suneil, and I totally agree with Suneil, is the fact that with the Perspectum study recently, with the UK Biobank, if we are really saying that 10% liver fat doubles your rate of risk of severe COVID and admission to hospital, then bringing this more to the fore where these devices are utilized throughout, that additional motivation, we haven’t learned the lessons from the previous sort of MERS, SARS, and H1N1, but maybe this time it is the opportunity where liver health is given a seat at the table for most of the comorbid conditions that are there, and monitoring everybody within health. Bringing it more to the fore makes it more affordable, and everybody should be having one. Every patient in every GP’s practice should have had some form of liver assessment. The more practical it is and the physical it is, the more it engages the patient.
Roger Green (00:53:52): Louise, one of the things I think we’re going to talk about here within the next few weeks goes to that, which is, what is the level of proof that you wind up needing around this link between viral pandemic and fatty liver, before society will mobilize in the way that it can when it really believes in something and doesn’t until then?
Louise Campbell (00:54:07): The only thing was going to say was something my husband said the other day, that if in 10 years time we hit another pandemic and have scanned and liver assessed everybody for fat and targeted the diabetes, the cardiovascular, the metabolic syndrome that this is all part of, and we get 100 million people healthier, we could have saved each country several hundred billion pounds. That’s exactly where we sit now. If we don’t do it, we’re back bailing countries out, and hundreds of billions of pounds. Or do we invest in health as a real, serious, growing concern? Particularly in countries with rapidly rising NAFLD and NASH. China, 29% of Chinese now have NAFLD, but those are the Chinese in the big cities that are more Westernized.
Louise Campbell (00:54:58): It’s a growing concern, but the question I suppose is, is anybody going to take any notice? Because we stigmatize the liver. We don’t talk about it enough in any comorbid condition. It’s not looked for in any guidelines, unless it’s a liver related guideline, so the medical world itself stigmatizes talking about the liver, how is the patient supposed to know or any individual to know that their liver really is one of the key organs in any disease process?
Roger Green (00:55:27): Well, I’ll add to that that one of the things I think we’re learning in the States this year is that people would like to deny the potential to get sick until they absolutely can’t deny it anymore. Part of the role of integrated leadership is to lead people to do better in situations like that. Part of my question becomes, at what point do you gain the momentum to get people who previously looked at the liver one way, to start to look at it differently so we can accomplish what you’re talking about? I don’t think it’s an easy top-down sell. I think, actually, the way the UK and the U.S., and how the two countries have handled COVID both prove it’s not an easy top-down sell, unless leadership is willing to do things that are difficult and unpopular in pursuit of health. Save that for another day, but that is a real challenge there, which is how do you build the social momentum to actually do it?
Roger Green (00:56:12): There are at least three good systems right now that are being developed and started to be promulgated and used for AI of the liver. PathAI, HistoIndex, Reveal Biosciences, and maybe more. I don’t mean to leave anybody out. Instead of talking about the nuances between the systems, I want to talk about why they matter. Really, they matter for two reasons. First of all, as we’ve discussed a few times on this podcast, and Stephen has an article soon to be published on this subject, one of the real problems with biopsy is actually reading the results.
Roger Green (00:56:41): There are really special things about biopsy. It is the cleanest way to figure out level of fibrosis. Fibrosis isn’t simply about how much collagen you have. It’s about the architecture and the form of the collagen within the liver. Suneil commented to me offline the other day that when you look at ballooning cells, you see capillary structures start to form around that, also architectural that suggest the potential for more serious liver disease pretty quickly. The ability to see a slice of liver gives us the ability not only to add things up and count them, but also to identify architectural patterns that could be precursors or something bad about to happen.
Roger Green (00:57:16): Problems are, to the naked eye, those things are amazingly, architectural patterns are hard to find, and using the scoring methods that we’ve got right now, we have tremendous variability in terms of what happens when two people read a biopsy slide, or even the same person who reads the same slide 12 months apart, depending upon the variable, correlation between the two readings is never higher than about .8, frequently as low as .5. In marketing research, if we did that with our clients, I’d get fired.
Roger Green (00:57:42): It means two things. Number one, it means as we discussed, we can’t really tell how good a lot of medications are simply at resolving issues of fibrosis, because we don’t know the degree to which the signal’s getting drowned in the noise of sloppy reading, number one. The second issue, which may, in some ways, be even more important than the first is, what happens if you try to predict these results? What most of us are taught at some point in life, even before we get out of high school, is that you have something called an independent variable and a dependent variable. And the goal of analysis is to figure out in some way, shape or form, what is it that makes the dependent variable change?
Roger Green (00:58:19): The question might be, so what can we really do to reduce fibrosis? How much of it is about different elements of a NASH score? How much of it is about other factors? What are the thresholds that have since become easier to do? For that to be right, you can either have been a hepatologist who’s been treating patients for dozens of years, tens of years, or a lot of years, and seen enough patients that you intuitively know what the patterns are. When Stephen talks about cirrhosis being reversible until you see portal damage, that’s quantitative, but that’s also what Stephen’s been treating for the last 20 years, or whatever.
Roger Green (00:58:51): The rest of it then becomes, “Can I quantitate that?” If you’ve got tremendous error around your dependent variable, then fundamentally, you’re trying to predict something really sloppy, and prediction is sloppy in the first place. Until we can tighten up our dependent variable measures, until we can be more consistent about knowing what we’re trying to predict, all the other tools that we’re talking about, better tests, better imaging will be limited, maybe more precisely telling us where we sit, but limited in terms of telling us what that means, and what’s the most important thing to do next because we don’t have a dependent variable measure that makes sense.
Roger Green (00:59:27): I’ve felt, for the longest time, that AI becomes a pivotal piece of this puzzle, whatever technology you use, because it really is the piece that will enable us to start to not only do a better job of identifying individual patients, but taking the aggregations of data that we’re talking about and lining them up against ultimate end results in the liver. We can do that right now with outcomes in other places, but outcomes of cardiovascular take, as Brian pointed out last week, tens of thousands of patients and maybe 20 years to sort out.
Roger Green (00:59:57): Well, if you’ve got liver disease, you don’t have 20 years to get this sorted out. You don’t even have two years to get this sorted out, so AI becomes a pivotal element in doing that. I don’t think we need to take time today and talk about what exactly it is makes for good AI or bad AI. All the AI that’s out there is good enough to make a significant difference, I believe. With that, comments, questions, thoughts? Stephen?
Stephen Harrison (01:00:19): As I alluded to in the last podcast, how do we improve the imperfect gold standard until we get to a non-invasive new standard? If you look at the data that’s been generated with AI off of liver biopsies, it’s made … This has all been done in the context of therapeutic drug development in the field of NASH, at least my familiarity with it has been that way. Each one of those, each time it’s been studied juxtaposed against semi-quantitative analysis, fully quantitative analysis using artificial intelligence has always provided a much more granular, detailed analysis of whether or not a drug was effective or not effective relative to fibrosis predominately.
Stephen Harrison (01:01:16): I think that we need to really focus on putting effort into developing that and augmenting the semi-quantitative analysis that pathologists do. I’m not saying replace pathologists, but we can augment and come alongside, and have a companion diagnostic that basically just expands on what they’ve shown. Until we can get to a non-invasive test, I think we really have to exploit that AI that’s out there.
Roger Green (01:01:51): I agree with that fully, Stephen. I also wonder whether if we had more confidence in the quality of the biopsy data that we get, whether in clinical trials or not, and that would be provided by better liquid testing, and better image testing. We can actually start to draw some conclusions that we can’t now, in advance of a perfect non-invasive or robust enough non-invasive test, or even to push us further down that path by telling us where it’s a good place to look more and more extensive, and a place to look more. Does that follow?
Stephen Harrison (01:02:23): It does, but even more specifically, we are learning about the pathogenesis of the disease even more so by applying AI. Now we know, for instance, that in certain zones of the liver where steatosis regresses, that’s also interestingly, where fibrosis regresses, where collagen bands begin to shrink and perivenular and pericellular fibrosis begins to dissipate. As we learn more about measuring and assessing ballooning degeneration, can we one day get to the point where we can say, it’s not a matter of going from some ballooning to no ballooning, well, what if you go from large clustering of balloon cells to maybe spotty balloon cells? Does that relate to long-term outcome? Maybe.
Stephen Harrison (01:03:18): What if we went from 60% ballooning to 20% ballooning? That’s something that can potentially be measured by AI that can’t be assessed semi-quantitatively by a pathologist’s eye. Maybe that can then be linked to outcomes. Maybe it could be linked to fibrosis, which is linked to an outcome. I think we have got to get better with the tools that are out there right now, and utilizing them to help us find data that is reproducible time-in and time-out, and is not subjected to this inter and intra-observer variability that plagues what we do today, and is making it challenging for the agency to accept a surrogate endpoint, in my opinion.
Roger Green (01:04:04): I completely agree with that. I was only going to add one more thought based on what you just said. Then Suneil and Louise comment, and then we’ll exit. If we were able to identify a region of the liver where reducing steatosis leads to reduction of collagen, and we then had a hand-held imaging device that could, say, as much as 5 or 10% of the liver in a series of scans that could be targeted, you would then be able to translate that biopsy knowledge into real time, on the ground diagnostics. The pieces are there to do that, or will be coming to do that.
Roger Green (01:04:39): At that point, I think, I totally agree with what you say. That’s how we get to accelerate approvals, because we really know what matters. We also then can put a device like that in Louise’s hands or Tawazun’s hands and go out and start looking at patients wholesale and figure out where is the risk really at a level far more significant and far easier to evaluate than anything we have right now. I think that’s where all this winds up coming together.
Roger Green (01:05:02): Everything we’re talking about today comes together where you can combine AI, which gives you a much better sense of where the target is, or what you’re trying to do, with better blood tests and better imaging that make it clearer how you could do that, and in the case of imaging, for less money so you can put it in the hands of more people. That would be a great direction, I think, to go in, and that’s my last thought for the day. Suneil, Louise, Stephen, any closing comments?
Louise Campbell (01:05:23): Just picking up from what you and Stephen were both saying there, I think we do have to take into consideration, when patients donate samples, whether it’s for clinical trials or to have their diagnosis confirmed, these are very valuable assets. If we can use AI to make the best judgment of those donated samples to get the best results, then AI and any future technology has to be used. Because if you are really going to convince a patient to have a liver biopsy, which they really don’t want, then it is our obligation to make sure that we get the best, accurate information out of that small sample that they are kind enough to donate at significant risk, at times.
Louise Campbell (01:06:10): I think by keeping advancing the technology, then we are doing patients a great service by really trying to take this beyond a physical needle that takes, what we consider a gold standard, and we’ve discussed a lot of times here isn’t the gold standard. Radiology biopsies, for example, are smaller than percutaneous biopsies performed by physicians, so you get more tissue on one than the other.
Louise Campbell (01:06:37): I think it is absolutely vital that we protect the integrity of every sample that a patient is kind enough to donate, particularly in clinical trials. Going back to the FDA’s change of heart, that undermines the samples that those patients gave if they are questioning the data that was taken. Again, we have an obligation to make sure that that is as robust as possible within the techniques that we’ve got now, without moving the goal posts. That would be my comment. It’s really to protect these. They’re absolutely vital, samples that they’re kind enough to give.
Roger Green (01:07:11): Thank you. Stephen and Suneil.
Suneil Hosmane (01:07:13): I love this topic. I think it’s fascinating. I think it’s going to uncover all kinds of wealth of information. For anybody who’s listening who’s seen a pathology slide of any tissue of any part of the body, you know you’ll see thousands of features. We talk about four features. Whenever we talk about NASH, we talk about steatosis, inflammation, ballooning, fibrosis. There’s thousands. Even the configuration like, Roger, to your point, the architecture of how these things are laid out and how they change over time.
Suneil Hosmane (01:07:46): Short answer is I think this approach is going to shed light not only on the disease, it’s also going to potentially give us information that’s going to lead to better drug development, maybe even better compound selection as well, being able to uncover how these things work. I think the approach is not purely a pathology approach. I think the same, to be honest with you, a similar approach was used by us to identify NIS4 just as an example. What’s interesting to your comment around blood vessels and stuff. Among many other things, microRNA-34a is an oncogene, it’s a suppressor, so as disease goes up and you see this marker begin to go up, the reason for an increased risk of ATC is not so decoupled from the rest of these processes.
Suneil Hosmane (01:08:37): I think having these type of hypotheses-free approaches where you’re digging into the data, you’re grabbing valuable insights, blood-based, pathology-based, maybe even imaging-based actually, to be honest with you, I think they’re going to be really remarkable and transform our understanding of the space. Of course, that will ultimately lead to better value and options for patients that are based on unique science. I think that’s what we’re missing is we’re missing additional color and insights coming so that we can learn as a field, and I think this is the right approach.
Roger Green (01:09:09): Thanks. Stephen?
Stephen Harrison (01:09:10): I don’t have anything additional to add. That’s good.
Roger Green (01:09:13): Good. That’s a first. Congratulations everybody, Stephen, for that assessment, and all four of us for putting it on the table to the point where you get to the place. That ends today’s episode. I want to thank the surfers, Stephen, Louise, Suneil, who were able to join us today. Once more, we talked about a lot of stuff today that should make patients interested and curious, and providers, and advocates, and developers, and everyone else send questions, send comments. We’d like to start to engage more often with people who have simple questions we can answer, complex questions we can answer, or would like to come on and talk with us about some perspectives they have about this podcast or any of the others.
Roger Green (01:09:49): Special thanks to engineer and podcaster, Frank Sasso, who makes us sound so good, social media master, Eric Rounds, who makes our stuff go round, and our editor and general social media support person, Ellyn Charap, who has been making this stuff work better. An extra special thanks to you, our subscribers and listeners, who’ve made us the fastest growing fatty liver podcast in the world. As we close, as we always do, what surprised you today? Louise, why don’t you go first.
Louise Campbell (01:10:14): I suppose what surprised me is how we’ve come so far with blood tests and these devices, but how far we still have to come, given the rate of diagnosis. I think that still surprises me.
Roger Green (01:10:29): Okay. Stephen?
Stephen Harrison (01:10:31): I agree. It is surprising that it has taken this long to come up with a non-invasive solution to identifying patients at risk for disease. I think it speaks to the heterogeneity of this disease and the challenges that exist in identifying these patients, because it’s not a one-size-fits-all. It’s a very complicated disease with very different presentations. Different presentations in different ethnic groups and different populations around the world. It is challenging, but we have come a long way and I’m looking forward to the next steps that we’re talking over the coming days, weeks, and years.
Roger Green (01:11:14): Thank you. Suneil?
Suneil Hosmane (01:11:15): I am surprised that, from a drug development standpoint, there isn’t more, at least even discussion of companion diagnostics or complementary diagnostics, or something to refine that kind of development strategy. Especially in light of this news coming from the complete response letter. Maybe that will change the field’s thinking.
Roger Green (01:11:37): I guess, what surprises me is I see this chasm. On the one side, I see that we are within a year or two years of having cost effective, non-invasive testing that will allow us to provide patients with information they can use to drive their own health better. On the other side of the chasm, we are waiting for different kinds of numbers from a disease paradigm that doesn’t work as well, and given the heterogeneity of the disease, we don’t know exactly what to say on the disease, on the illness side.
Roger Green (01:12:08): My hope is that we can figure out how to bridge that chasm and understand that imperfect information will still help patients to take much better care of themselves, if given to them properly and supported properly, and that these tools will, at the same time, enable us to start coring into the heterogeneity of this disease, understand it better.
Roger Green (01:12:25): I want to thank you guys. I think this has been a spectacularly interesting session. Every time we get to the end of one of these, I’m surprised by how much I’ve learned and how interesting it’s been. We will be back on Thursday, July 16th with episode 14. Don’t know what it will be yet but I’m guessing it will either have to deal with patient empowerment or selecting winners and losers in some shape or form. Until then, everybody stay safe. That may take a lot of work depending on where you live in the world, but stay safe. Surf on. We’ll see you next week.
Speaker 2 (01:12:51): You’ve been listening to Surfing the NASH Tsunami. Send in your questions to surfingnash.com and our panelists will spend the first five minutes of next week’s episode answering your questions. Visit us online today, surfingnash.com.