This is my disclosure slide. And this is the case we were provided. It's a 33 year old man known to have PSC for six years. Then something happened. He developed low grade fevers, scleral ectoris, and poor apatitis. And then we are given three biomarkers. What I will do in this talk, I will talk you through some reflections on those items in this patient story rather than necessarily providing you with a comprehensive overview. We'll start with the first point here. We have a PSC diagnosis and we have a high serum IgG4. So how do we reflect upon those two items? We start with a diagnosis. Sclerosingulangitis is in many ways a final representative of a common pathway of chronic biliary injury. And there is a variety of etiologies that can give rise to these changes. And only upon exclusion of these known etiologies you can establish the diagnosis of primary sclerosingulangitis. And you will see amongst this various vascular, inflammatory, infectious etiologies also IgG4 related systemic disease. This wide variability in secondary sclerosingulangitis etiologies have led some to claim that PSC as such may be an equally heterogeneous entity. I personally don't necessarily reflect upon it that way. I think at least there is one major subgroup of patients with concomitant inflammation in their bowel. Basically the liver, biliary, and bowel disease entity which is fairly homogeneous when it comes to pathophysiology. But of course can have variable disease expression. An important feature of this entity is the development and the risk of ulangiocarcinoma and also the higher risk of colonic cancer that we see in these patients. Then to the IgG4 point, this is a very simple but to me pretty clear illustration of two different categories of PSC patients that we meet. Most of our PSC patients they have low IgG4 levels but there is a sub-fraction, 10, 15% that actually have higher IgG4 levels. Normally however they're not as high as was given in this case. They're typically in the range of 200, 300, 400, maybe up to 500. But going beyond 500 you would go into really the exceptional cases and normally you are actually quite straightforward into the domain of IgG4 associated cholangitis. The problem also is that in cholangiocarcinoma you observe this same kind of headache zone of IgG4 elevations. Some patients have these 300, 400 type of IgG4 levels and really to differentiate based on the serum profiles only versus IgG4 disease can be exceedingly difficult. Also as you can see some of the patients with true IgG4 disease, they actually have normal IgG4 serum levels. So there may be a means of sophisticating these considerations. You can do IgG4, IgG1 levels. There are also now PCR detections of B cell clones specific to IgG4 associated disease. But the most important message I think taken from this table is that IgG4 alone is a poor biomarker. You really need to see it in the context for the IgG4 disease and other organ involvement for PSC with IBD. And of course these considerations are of major importance to the patient because it determines its outcome. It is management. Should he have steroids? Should he have a Whipple? Those are quite, I would say, diametral considerations. If you look at it from the IgG4 perspective, it somehow eludes simplicity. You can make these nice overviews of type one, type two, type three, type four IgG4 disease, very sophisticated illustrations of how the cholangiogram may look. But in reality, most of these features are nonspecific and may also occur in the context of PSC and cholangiocarcinoma. And in the end, it's a compound syndromic consideration. And you have to consider whether or not to give a trial attempt of a corticosteroids based on multiple, at least two features of considerations. And we will make mistakes. Sometimes you will miss an IgG4 patient because the serum IgG4 is not really elevated. You will think that you have an IgG4 patient whilst it's actually a PSC patient with slightly elevated IgG4 level and it will not respond to your corticosteroids. And every now and then, unfortunately still, our hepatobiliary surgeons will come to us with a Whipple resect that actually shows that there was an IgG4 disease. But considering the features that I've now illustrated to you, I think that those errors can be minimized. And this is really where we should be heading. If we move on to the next consideration of this case, six years, it's said. Is that a long time? Is it a short time? And then we have an ALFOSC level of 162. Is that high? Is it low? It's about 1.5, the upper limit of normal. These are four Kaplan-Meier plots. One showing our OSLO patients. The median survival time, somewhere around 12, 11, 12 years, which is probably due to a referral bias. We're a tertiary referral center doing transplants. And if you look to the lower left, you have the Amsterdam series, which is a population-based series, where you have a median overall transplant-free survival of 20, 22 years. To the right, you can see the ALFOSC stratification on 1.5 from the Oxford patients. And actually, then it also brings forward this problem that we have. Okay, so this patient, basically exactly 1.5 times the upper limit of normal. Does he go with the upper, the good prognosis category, or does he go with the poor prognosis category? We cannot really tell on ALFOSC alone. Also, if we consider IgG4 as a biomarker in PSC patients, not IgG4-associated scleromyelitis patients, but PSC patients with elevated IgG4 levels. IgG4 also serves as a biomarker for poor prognosis. These are the Mayo patients. Really, the overall theme from all these Kaplan-Meier plots is that PSC is a progressive disease. With time, you have this subtle inflammation, you have the cholestasis, you develop fibrosis, and ultimately end-stage liver disease. You have complications, clinical events over the way, you have strictures, cholangitis, and ultimately you will need a liver transplantation. Overlying these slowly progressive features given by the statistics, at the individual patient level, there is a lot of fluctuations. ALFOSC biochemistry goes up, goes down, there are SHUBS probably, you have IBD activity going up and down, and really making our assessments and establishing of prognostic modeling quite hard. This is probably reflected also in an overview that you could put forward like this, where you have quite a variable number of features brought forward in terms of prognostication of PSC. And one very imminent feature of this overview is that many of these parameters are actually parameters of advanced liver disease. You have albumin, INR, you have the MELD score, and there are features of advanced cholestasis. Bilirubin is almost all over. And when you try to take these prognostic scores one step further into trying to make consideration as to disease severity, so okay, you have those high bilirubins, low albumin patients, but did they have a severe PSC or did they just have PSC for a very, very, very long time until you established your score? And I think that these two diagrams now shown on the slide are quite interesting. The one to the left is an expert Delphi consensus round asking PSC experts in the international group which biomarker they think somehow would be useful in predicting PSC activity in clinical trials. The other one is a machine learning approach to a clinical data set, basically asking on prognosis of Mayo Clinic patients. And as you can see, the expert panel, bilirubin as a marker of advanced cholestasis doesn't really score very highly, whereas in the machine learning as expected and as seen from these more historical scores, it shows excellent utility. If you switch around and look at ALKFOS, the experts are very enthusiastic about ALKFOS. However, the machine puts less emphasis into the ALKFOS. If you refine this consideration, it becomes much more complicated than it currently is in PBC. These are the UK data. And what they propose, which I think is an interesting thought, is that in the imminent events, of course you would be looking at markers of advanced liver disease, but in the long-term run, up to 10 years and beyond, maybe ALKFOS and other markers of true disease activity become more and more relatively important. Of course also, again, as you can see from these formulas, it occurs in a context. It's like IgG4. You don't consider ALKFOS in isolation. And I think two areas where there has been a lot of work and where we will see a lot of emerging work as digital image analysis evolves. We have artificial intelligence, et cetera. But these are human interpretations. These are the French patient series on MRI, MRCs. And I think that the interesting feature of the MRI in prognostication in PSE is actually that it's not really much of a utility when it comes to the biliary structure. It's really hard to find good biliary prognosticators. But in the parenchyma, you actually have the strongest prognostic signal. And that's probably also the reason why you see this prognostic utility from the biopsies. Again, this parenchyma signal, these are the Nakanuma stratifications in an international panel of liver biopsy assessments in PSE. So there is this parenchyma signal, and this is also what you can see and what you can detect using elastography. This is the transient elastography data from Paris. To the left, you have the single point measurement. To the right, you have the change over time. And now there is also, actually, as of November 1st, the MR elastography data from the Mayo Clinic out, and they show somewhat the same features. And I think this is also where much of this field is currently moving. It makes a lot of sense to look into the fibrosis domain in PSE. It's truly a fibrotic disease. This is the principal lesion. There is a dynamic to this fibrosis, as we heard about. And you have the fibrosis occurring at the liver level, and you have the fibrosis occurring at the biliary level. Ultimately, then, also, these markers show promising utility. And I think that the next step, and really where the field and the community is now standing, is to bring these biomarkers together with the imaging data, together with the clinical prognostic scores, integrate them and simplify them into tools that we can actually use in the clinic. So far, it's mostly retrospective and not really prospective and validated data. Final point, something happened to this guy. There was something that was not really there for those six years. He got this jaundice, and what is the consideration in this context? I already mentioned the high risk of Kullerianger carcinoma. There is also this high risk of colonic neoplasia, as illustrated here from the Amsterdam series. And you can see that this risk continues to increase, and this we also see in the international PSE cohort, even beyond the first 10 years of diagnosis. And then the question is, is this then an IgG4 stricture in this patient with the high levels of IgG4? Is it the benign stricture? Is it the cholangitis? We don't know, but you should worry. I think that's the key message on this happening. This happening is a very fundamental indicator, and almost like a biomarker in itself, I would say, that should prompt you to think about malignancy in this patient. Then, of course, you have the problem of ruling in or ruling out Kullerianger carcinoma, which is exceedingly and can be exceedingly difficult in the early stages, particularly when you are at the biliary dysplasia stage. But you move through the tools that you have. You have your biochemical parameters. You have your imaging tools. And then there is this opportunity of ensuring a tissue diagnosis. All these tools, somehow imperfect. We will be back to that in a minute. But nevertheless, the question arises then, so should we use them? Should we use them for screening? For screening for colorectal neoplasia, it's super simple. We do colonoscopy. And this works. This is shown here in this Amsterdam series that there is lower colorectal cancer-related mortality when you have the patients under colorectal screening, obviously. And then the question comes in PSC and Kullerianger carcinoma. We have those imperfect tools, imperfect biomarkers, imperfect imaging. Should we still do surveillance? And this is an interesting paper from the Mayo group, again, where they look back at their patient series to 1995. They've had some form of imaging, biochemical surveillance to their patients. And they actually see that there is a utility when it comes to decreasing outcome and colorectal cancer, or Kullerianger carcinoma-related deaths in the patients who were under surveillance. So I think that the message from this figure is that even though our tools are imperfect, they are not good, our algorithms are not good, but we should still not do nothing. I think that this alternative is really not what we want to stay with. So what about these tools then, these imperfect tools? This is CA99. It's a feature of the glycolipid layer of the epithelium, both in the biliary tract and otherwise, other epithelial surfaces. And of course, this can increase in the context of infection, of inflammation, in context of IgG4 disease itself. But it can also vary depending on the genotype of the individual. Depending on the foot three, foot two genotype of the individual, you can have either a total, almost non-secreter of CA99, or a very high producer of CA99. And that means that those levels that we're talking about here, what's the high CA99 level? Is it 70, 100, 150, 200? They become very difficult. And I think, again, you have to put the CA99, like we did for ALKFOS, like we did for IgG4, into a context, into a clinical reflection. I think isolated measurements are not really good. The context, of course, in which we try, particularly I would say in Scandinavia, in the U.S., is to put the considerations into the context of tissue diagnosis. We're very eagerly, at least in Scandinavia, trying to get some sample of the biliary epithelium. We know that the sensitivity is low. We know, even in combination with fish, that the accuracy is low. But bringing this together with the biomarkers, together with the clinical considerations, we actually do achieve some help in establishing management algorithms for the patients. Also, there is an increasing drift, and I think this is important to mention, in towards advanced endoscopy in the form of ulangioscopy. We're doing more and more ulangioscopies, also allowing us to get a tissue diagnosis in those challenging cases of, is it dysplasia, is it ulangiocarcinoma, or is it just a regular stricture? And it helps, at least when it comes to the considerations for the preliminary data. Finally, of course, as we have heard alluded to in other talks in this post-graduate course, also in the omics space, explorations are being made, particularly when it comes to this early detection of ulangiocarcinoma. You have proteomics, you have DNA technologies, you have metabolomics, et cetera, et cetera. And I think, as also computational capacities, machine learning, digital integration of all these data, with the clinical data, with the imaging data, with those imperfect scores that we already have, I think that we will see some enhancement of the quality by which we manage our PSE patients in the very near future. So far, we're stuck with something very simple when it comes to the follow-up of PSE patients, but we should see them. I think that's also the lesson taken from the Mayo paper on the surveillance. We should see the patient regularly. They should be seen by a specialist at least once or twice a year, and we should do our imperfect tests, and we should listen to them when they tell us that something has happened, something has changed. And then we should worry. I think that's the first key take-home message. I think we have some reflections here on the individual markers. They need to be considered in context. It goes for IgG4, it goes for CA99, and it goes definitely for ALKFOS. We do see an increasing application of imaging and allostrography, and also non-invasive fibrosis markers, seromarkers, in prognostication of PSE. I would advocate some aggressivity in the securing of a tissue diagnosis when it comes to early cholangiocellular dysplasia detection, and I think that where this field will be moving in the very, very near future, we will see the synthesis of multiple diagnostic modalities into what you could call big biomarkers with omics data, clinical data, imaging data, all coming together, helping us in these challenging situations. Thank you very much.

primary sclerosing cholangitis

biomarkers

IgG4 levels

ALKFOS levels

CA99 levels