Dr. Lo raises some important issues regarding treatment of patients with ascites and about practice guidelines and their levels of evidence. Having been the author of the very first practice guideline commissioned by the American Association for the Study of Liver Diseases (AASLD) as well as a member of the Practice Guideline Committee of the AASLD, I have a unique perspective on the process of writing, editing, and reviewing guidelines.1 An important issue that Dr. Lo needs to realize is that an author cannot simply make revisions of a guideline on an as-needed basis whenever there is a request for a revised (1) statement or (2) recommendation or (3) level of evidence. The guidelines are considered for an update yearly but are actually updated when the author(s) and the guideline committee agree that there is enough new information to warrant a revision. The process of revision can take more than 2 years. I was invited to revise my guideline in January of 2007, but the guideline was not published until June of 2009. The committee review process, board and society review process, and the possible external review all take considerable time. Frequently, the author(s) would like a very long comprehensive document that attempts to cover all scenarios and contingencies, but the board seems to prefer a much shorter document.
Also, there are AASLD guidelines for many topics now, including hepatitis B and liver transplantation. Where to put specific statements regarding patients with hepatitis B and ascites or hepatitis B and transplantation could be debated. Although I agree with Dr. Lo that patients with ascites and cirrhosis due to hepatitis B should receive antiviral treatment (other than interferon), that recommendation (#24) is actually made in the hepatitis B guideline.2 It will be a consideration to also include that statement in the next iteration of the ascites guideline. Dr. Lo would also prefer that patients with decompensated hepatitis B cirrhosis only be considered for transplantation if their liver failure is “…not amenable to control” after antiviral treatment. This is self-evident. The astute hepatologist would not attempt to perform transplantation on a patient who was asymptomatic with a low Model for End-Stage Liver Disease (MELD) score, perhaps taking only one pill a day—that is, the antiviral. The MELD score, which usually improves dramatically with antiviral therapy, would actually decrease to a level that would prevent the patient from receiving a liver transplant in the United States, unless hepatocellular carcinoma developed. There are perhaps an infinite number of specific situations in which recommendations for consideration or no consideration of transplantation could be made. These may be better placed in the guideline on transplantation. Page space limits the number of specific scenarios that can be addressed in a guideline.
Dr. Lo also is of the opinion that vasoconstrictor treatment of type 1 hepatorenal syndrome is Class I, level A rather than Class IIa, level B, quoting the same publications that I quoted. The author of the guideline actually does not have the final say on what class and level is given for a recommendation. The committee has the final say with the input and approval of multiple other entities, including the AASLD board and other societies. In this instance, the class of evidence issue boils down to semantics. Class I requires a treatment to be “beneficial, useful, and effective”. Vasoconstrictor treatment is frequently effective in improving renal function, but there is no survival advantage. Vasoconstrictor treatment would be considered Class I regarding renal function but Class IIa regarding survival. It will be a consideration to change the level of evidence from level B to level A in the next iteration of the guideline in view of the growing database on vasoconstrictor therapy of hepatorenal syndrome.
Drs. Gerbes and Jungst would like to know why cholesterol is listed as “unhelpful” in table 3 of the ascites guideline. Actually, many other ascitic fluid tests have been proposed as helpful in separating ascites due to cirrhosis from ascites due to malignancy—including fibronectin, ferritin, interleukin-2, beta-human chorionic gonadotropin, glycosaminoglycans, amino acid gradients, atrial natriuretic factor, bile acids, alpha-1 antitrypsin, sialic acid, soluble tumor necrosis factor, etc. Most of the abstracts on these tests never made it to full publication. I have written about these “humoral tests of malignancy” repeatedly.3, 4
Most of these small series compare results from patients with cirrhosis to those of patients with peritoneal carcinomatosis. Most protein-based tests on fluid from these two wildly different sources show a near complete separation of data. The problem is that when a clinician is faced with a patient with ascites, that patient could have any of many, many causes—not simply cirrhosis versus peritoneal carcinomatosis. Some patients with cirrhosis also have peritoneal carcinomatosis. Some patients with cirrhosis also have hepatocellular carcinoma or spontaneous bacterial peritonitis. Some patients with cancer and ascites have massive liver metastases, peritoneal carcinomatosis and massive liver metastases, malignant chylous ascites, malignant Budd-Chiari syndrome, etc.5 Patients with hepatocellular carcinoma usually also have cirrhosis; they almost never have peritoneal carcinomatosis. These subcategories of patients were usually excluded or ignored in these small series.
In contrast, the study that validated the serum-ascites albumin gradient (SAAG) included 901 samples and “all comers” with ascites, with a serious attempt to evaluate all causes including multiple causes for ascites formation.6 Even the SAAG does not tell all there is to know about the cause of ascites; it only separates portal hypertension from etiologies not related to portal hypertension. Some patients have both portal hypertension and non-portal-hypertension–related etiologies, and their serum-ascites albumin gradient is in the high range. Ascitic fluid analysis is complicated because of the diverse and sometimes multiple etiologies.
The only test to be definitive in confirming peritoneal carcinomatosis is the cytology analysis.5 This is an excellent test when processed properly and positively. Most of these humoral tests of malignancy are falsely positive in patients with high-protein, low-SAAG ascites,3, 4 such as ascites due solely to tuberculosis or pancreatitis. The various series of these humoral tests of malignancy included almost no patients with tuberculous peritonitis or pancreatitis, and almost no patients with hepatocellular carcinoma. Because hepatocellular carcinoma seldom causes peritoneal carcinomatosis, the usual ascitic fluid analysis resembles that of cirrhosis, only there are more bloody samples in patients with hepatocellular carcinoma.5, 6
A large consecutive series of “all comers” with ascites is required to validate a new test. Small series of selected patients lead to more confusion than clarity. Actually, more humoral tests of malignancy should be listed in the unhelpful column of table 3, but page space is limited.