All right. Thank you. Well, I'd like to thank the chairs and the organizers for the invitation. As always, it's an honor to present at this meeting, and I look forward to it as often as I can. I also look forward to controversies, sessions. They're fun. I can say things that are controversial, and I'm not going to shy away from that. I do plan on at least backing up some of the stuff I say with data, so you can look forward to that. I can't imagine any conflicts of interest in this topic. Okay. Let's start with a case here. Any of us that do endoscopy, especially endoscopy with a large group, you get other people's patients put into your schedule quite a bit just for convenience. And here's a case of mine, 3.30 p.m. outpatient, 49-year-old man with hepatitis C-related cirrhosis, workup for liver transplants ongoing. He has a history of esophageal varices and unable to tolerate nonselective beta blockers. No other major medical problems. He's put in your template late in the day. You've never met him before. And here's the picture of you as you're deciding what to do about this patient. There's no labs in the EMR. You don't really know the patient. You look back and see that there were some labs from a month ago. He had 55,000 platelet count then. Hemoglobin was at eight. INR was 1.9. Normal renal function meld 23. This is exactly the kind of case you want to see in your late in the day open access cases, right? So what do you do? And what do you do about this particular patient? A, do you cancel the case? It's too dangerous due to bleeding risk because of the coagulation disorders that this patient has. B, you can order the labs, check your updated labs, wait around, transfuse platelets and FFP, I guess, recheck labs, and by the time you get the numbers back, it'll be midnight and you're going to do that in the middle of the night. Is that what you're going to do? C, you could reschedule and just give it back to one of your other colleagues and that happens a lot. Or D, you could just do the case. And I hope maybe I can convince you on what the right thing to do is later. We're going to talk about why liver patients bleed and sometimes why they clot or the rebalance of hemostasis as it should be thought of. If we want to talk about preventing bleeding or changing the risk of bleeding for a procedure, we should at least define what the actual risk is. And that's much harder than it actually would seem. But we'll talk about that. I'm going to rail on the use and the misuse of INR for bleeding risk. And then finally, maybe now we finally have some better predictors of bleeding than just INR. So let's go through this. First of all, briefly, the rebalance of coagulation is a relatively new concept. Over the last 10 years, we've understood that yes, the INR is abnormal in cirrhosis patients. Yes, the platelet count is low in cirrhosis patients. But in this paper from Armando Tripati now in the New England Journal now eight years ago, if you can see things imbalanced on one column, there's a corresponding rebalance in another column. For instance, the low platelet count in cirrhosis patients is balanced by high von Willebrand factors. We check platelet counts all the time in the clinics. We don't check von Willebrand factor antigen very often in clinics. Similarly, the vitamin K deficient coagulation proteins are rebalanced by deficiencies in antithrombin and protein C. We don't check those. We check the INR. So we're looking at just one side of the puzzle in cirrhosis patients. This is much different than patients on warfarin or Coumadin. An INR in a patient who is taking warfarin is indicative likely of bleeding risk, whereas in cirrhosis, we have these rebalancing things. Now, don't get me wrong. Patients with liver disease, as you all know and we just saw in the past 25 minutes, bleed. They also clot. When I get asked to speak on coagulopathy and liver disease, I speak on clotting a lot, portal vein thrombosis. We all struggle with those issues. Today, we're going to focus on the clot breakdown tendency or the bleeding tendency of liver disease patients. But remember that there's a balance and most patients walking around with liver disease are in balance. They're not bleeding or actively clotting because of the pathophysiology we just discussed. So let's run through some of the common procedures that patients are exposed to with liver disease when we see them. First, paracentesis. Paracentesis in the decompensated liver disease patients is probably the most common procedure performed. This study is indicative of the literature out there. This is from the Mayo Clinic. 1,100 therapeutic paracentesis were performed on these patients. As you can see in the table on the right, their coagulation parameters to traditional coagulation parameters were all over the board. Most patients had platelet counts less than 60 and the great majority had INRs over 2. And yet, in those 1,100 therapeutic paras, no procedural complications requiring hospitalization. Only one episode of a bloody tap, the patient was discharged that day. Four episodes of fluid leakage and no clinically significant bleeding in any of the procedures. The literature out there is pretty rife with studies like this. And our society, the ASLD, actually has a recommendation on paracentesis. This is quoted directly from the practice guideline. Bleeding is sufficiently uncommon. Therefore, the routine prophylactic use of FFP or platelets before paracentesis is not recommended. No thresholds listed. Don't check before you do the paras. If you're giving plasma before paracentesis, you're not following the practice guidelines for routine use. Let's talk about liver biopsy. Liver biopsies, this is what we all fear. It's an uncontained space and people can put a lot of blood in their peritoneum. The last thing we really want to see in our liver biopsy specimen is in the middle there. A liver biopsy along with a third of the artery wall of a large artery next to it. So we want to avoid bleeding. We have some guidance on bleeding in percutaneous liver biopsy as well. This is from the HALT-C study. These were cirrhosis patients. There are many people in this room that may have participated in the HALT-C study. There were 2,700 or more liver biopsies performed on cirrhosis patients. The bleeding rate from these liver biopsies was 0.6% of the entire sample. So there were 15 bleeds in this study. The table on the right outlines what the platelet counts were on those patients. The conclusion from this paper was that the highest risk for bleeding after percutaneous liver biopsy was in platelet counts of less than 60,000. While that is true on a percentage-wise, note that 11 of the 15 bleeds happened in patients whose platelet count was above 60,000. So perhaps platelet count isn't a very good predictor of bleeding. In fact, 8 of the bleeds, more than half, occurred in patients with what we would consider a normal platelet count, above 100,000 in cirrhosis patients. So that 60,000 made it in the practice guidelines for percutaneous liver biopsy. However, my interpretation of this is it's not a very good predictor of bleeding risk in liver biopsy. And there are some abstracts on this topic at this meeting. What about endoscopy? We as hepatologists and gastroenterologists do a lot of endoscopy on these patients. This is a study out of MD Anderson looking at severe thrombocytopenia and endoscopic procedures published five years ago now. 617 endoscopic procedures, 90 of them colonoscopies. And the gist of this, the findings of this article were that the platelet count before endoscopy was not predictive of bleeding after the endoscopy. In fact, either transfusion of fresh frozen plasma or platelet counts prior to the procedure were also not predictive of bleeding. So not only was the count before endoscopy not predictive, but the interventions that were performed were also not predictive of bleeding. In fact, this group, a high volume endoscopy practice, said that the risk of interventional bleeding is minimally increased and typically minor and easily controlled endoscopically. So I think we fear endoscopy in these patients more than we probably should. The literature supports that. Remember, as we just heard, that esophageal variceal bleeding is a mechanical issue. It's a pressure issue, right? Remember, it's the wall thickness, the size of the varix. It's not really a reflection of coagulation status. Variceal bleeding is a pressure phenomenon. That's why we treat with beta blockers and so forth. So if you're worried so much about coagulation disorders, you shouldn't be worried so much about esophageal variceal bleeding. So treat the pressure. Don't worry about the coagulation disorder. And we'll talk about that in a little while. I'm not the only one to make this statement. This is, remember, the Choosing Wisely program. This is also from the ASLD along with the ABIM. This was a flyer that was designed to be posted in clinics so patients could read them in the waiting area. And look at number five. Don't routinely transfuse fresh frozen plasma and platelets prior to abdominal paracentesis or endoscopic variceal band ligation. So I think if you're routinely transfusing products prior to elective outpatient band ligation, you're not particularly following what this society has stated would be the standard of care. So keep that in mind. Lastly, for procedures, colonoscopy. There's not a ton of data on therapeutic colonoscopy. But this particular study looked at polypectomies in child's A and B cirrhosis. And the conclusions from this study were essentially that the larger the polyp, the higher risk of bleeding. And that's true in the general population. That's not unique for cirrhosis patients. In fact, in this study, there was only a 3% bleeding rate after polypectomy, which is also similar to the general population. And you're getting the gist of my argument here when you read the last bullet statement. Platelet count, INR, child PU score did not significantly differ between bleeding and non-bleeding. So don't fear the liver disease patient. I know I'm preaching to the choir here when you are in endoscopy. I think we have some things that we can fall back on. Now, you would think the Endoscopy Society, at least the American Endoscopy Society, would give us guidance on this, seeing how there's data out there for this. The top two guidelines that I have posted the banner for here don't contain the word cirrhosis in either of those articles. They consider an elevated INR due to warfarin exactly the same as an elevated INR due to liver disease. And as we just talked about, those are two completely different pathophysiology. So we're let down by our Endoscopy Society, anyone who's a member of the American Society. The final guideline talks, I think there are two sentences about a platelet count in managing esophageal varices. And they, in fact, say 20,000 is safe for endoscopy. So if you're getting numbers quoted to you from other sources, I'd love to see those sources. They're really not particularly pertinent to the things that we do day to day. We actually have started a multicenter prospective trial looking at bleeding rates after common hospitalized procedures or patients who are hospitalized. We're following them forward. And there are many centers throughout the world that are participating in this. And we're about to kick this off. It is supported by the International Societies Club. So anyone interested in enrolling patients in this prospective non-interventional trial, let me know. Okay, in the last few minutes, let's talk about INR. Don't get me wrong, okay, INR is a great predictor of mortality in chronic liver disease. It's in the MELD score. It's in Child-Pew. We allocate liver transplant by MELD score, which is the INR. It's very good at indicating the protein synthesis dysfunction in chronic liver disease. But it's not very good at predicting bleeding in liver disease patients. And there are now more than 50 studies out there. And I stopped updating this slide a while back. There are more than 50 studies showing that INR elevation in liver disease patients does not give you helpful information for predicting bleeding prior to procedures. And you can see all the lists of different styles of procedures there. You can look those up yourself. They're all over the place. Yet we get this all the time from radiology, from surgeons, even friendly fire from our colleagues on occasion. Please fix the INR. First of all, if someone knows how to fix the INR in a liver disease patient, please call me. I'm interested to hear how you can do that. The radiologists want the INR normal for a paracentesis, for instance. So we have this fight. And it's a fight that continues every year in July. It's even more. I will tell you that my colleague, Steve Caldwell, reminded me of this paper that I wasn't familiar with from 1974. I had to go print it up at the library. But there is a linear correlation between the blood volume and plasma volume infused and portal pressures. It makes perfect sense. If you're going to give a lot of volume by vein, the portal pressures are going to increase. You're just increasing blood volume. And this is proven in the literature. So the more FFP you give, the more portal hypertension you're going to induce for the lifespan of that infusion you're giving. We see it in the units all the time. So if you're giving plasma before you're trying to ban varices, you're maybe not quite working on the same thought process that pathophysiology should lead you towards. So I don't know why we haven't embraced a restrictive transfusion strategy for things like procedures. The restrictive packed cell transfusion strategy has been validated in large multicenter randomized trials. We keep hemoglobins at seven versus nine, and there's a survival benefit. We don't have the randomized trials in INR and plasma yet, but we're working on that. But I think that's coming in the future. So perhaps controversial, perhaps not. So my endoscopy or radiology colleagues say, all right, great, we won't follow the INR. What do we check? Give me something to check. They want to check. So we need something to check. And in the past, we haven't had a lot of answers. I think we are getting more answers now. Platelets, I think, are important. And this was a landmark paper now 13 years ago from Armando Tripati where he found that cirrhosis patients using cirrhosis plasma as a testing vehicle could generate an adequate clot, at least within two standard deviations of the healthy normal population in platelet counts between 50 and 60,000. So I think that 50,000 mark for platelets has some physiologic benefit. This was an in vitro study. And we now have agents that we can use to enhance platelet counts. This paper from Gastro last year now looked at avatrombopag, but there are several others out there. Now, these studies were designed to show an ability to increase platelet counts above 50,000. The end point for all of these trials is an ability to raise platelet count above 50,000. They did not set the end points for bleeding. So we don't actually know if pushing the platelet count above 50,000 prevents bleeding. We just know whether these drugs are able to increase the platelet count above 50,000. So there's a couple leaps of faith in using these. However, if your goal is to increase the platelet count above 50,000 in somewhere between two-thirds and 88% of patients, you can do it with these medications. But to stress the point, directly quoted from the article, overall across the two studies, the incidence of bleeding events was comparable between the study group and the placebo group and were low in both platelet count cohorts. So I'm not so sure that getting the platelet count above 50,000 is a laudable goal, but we need to look for better predictors of bleeding. But we do have these drugs to help us at least reach those platelet counts that are written in some of the guidelines. Notice there were two portal vein thromboses in these studies, both in the treatment group but not statistically significant. I think fibrinogen is useful. We check fibrinogen routinely in our patients, especially our actively bleeding patients, and this paper from Vienna supports that practice. They found briefly in the interest of time that fibrinogens less than 100 and a platelet count less than 100,000 in critically ill cirrhosis patients were predictive of future bleeding. They didn't study whether intervening on the fibrinogen or the platelet counts actually helped prevent those bleeding events either, but in the actively bleeding patients especially, we routinely push fibrinogen using cryoprecipitate up above 100, sometimes to 150, and we try and get the platelet counts up in the actively bleeding patients. We do fibrinogen in patients undergoing high-risk procedures. We will check fibrinogen pre-procedure and supplement in the perioperative time frame if necessary. Finally, the use of whole blood viscoelastic testing is expanding. There are several abstracts out there on this and several papers. Using things like Rotam or thromboelastography, you can measure the whole blood clotting response for a patient. You get a nice curve that shows not only the speed of formation of clot but the maximum amplitude and actually the breakdown of clot. This is a paper out of Italy published three years ago now looking at using thromboelastography to guide transfusion before procedures. There are a lot of issues with this study, but it's a nice proof of concept. Using very specific characteristics and measurements from the thromboelastogram, these investigators essentially randomized people into two groups. Using TAG to predict the need for transfusion prior to procedure or just transfusing everyone before these procedures. You can see in the left column of the table, the TAG group got only 16%. 16.7 needed transfusion based on the thromboelastogram versus 100% in the standard of care group. You can see the volume of medications administered. 4,000 milliliters of plasma for the TAG group. 11,000 milliliters in the standard of care. Six units platelets versus 78,000 units of platelets. 78 units of platelets rather. Note that neither the INR nor the platelets really changed in the transfusion group. So again, we get back to this ability to change things. And also note that there was no procedure-related bleeding in the TAG group. Although there was one in the standard of care group and one transfusion reaction. So there's way more study to do. But I think we have some information now to use these devices going forward. So back to the case in the last 30 seconds. You're going to cancel the case because the bleeding risk is high? Well, we don't think the bleeding risk is that high for this patient. And the annual bleeding risk for varices is high as we know. If you order labs, you're not going to really help yourself in prediction of bleeding. You're going to delay things. And, you know, perhaps a Rotem prior or a TAG prior to the procedure might have helped you. Rescheduling into your colleague's schedule is going to get you in trouble in many other ways. So I did the case. I do this routinely. And that's how this ends. So I'll skip, in the interest of time, my key takeaways. Thank you very much. Thank you.

liver disease

procedures

bleeding risk

cirrhosis patients

viscoelastic testing

evidence-based approach