Statistics from Altmetric.com
The recent letter by Jansen et al 1 presenting an algorithm to rule out clinically significant portal hypertension (CSPH) by combining shear-wave elastography of liver (L-SWE) and spleen (S-SWE) is an important contribution to the increasing knowledge regarding non-invasive assessment of patients with compensated advanced chronic liver disease (cACLD). Baveno VI recommendations focused on transient elastography as a tool to rule in CSPH in patients with the authors cACLD of viral etiology,2 ,3 but there is no doubt that other related elastography techniques should work similarly. Also, recent UK guidelines on variceal bleeding introduced elastography for variceal screening and diagnosis.4
In their study, the authors concluded that combining L-SWE <16 kPa with S-SWE <26.6 kPa showed a very high sensitivity (98.6%), with a remarkable negative predictive value for CSPH (96.3%). However, we have a few comments about the design and interpretation of results that we hope may enhance the transferability of these excellent results to other cohorts and settings. The first point is in regard to the selection of patients. The authors included many patients in Child–Pugh class B and C (37%) and decompensated patients with ascites (43%). These patients have by definition CSPH and they do not need non-invasive test for this diagnosis. Thus, we feel that inclusion of these patients should be avoided in diagnostic studies for CSPH. Indeed, the prevalence of CSPH in this cohort (ie, the pre-test probability) is very high (66%). The classification rules obtained from such a population might not be applicable to the population of interest, which are patients with cACLD, in which the pre-test probability is expected to be much lower. In our view, the more uncertain the condition of interest (here, CSPH), the greater the expected value of non-invasive tests. It is in this population of cACLD or early cirrhosis with a relatively low risk of varices and CSPH that the Baveno VI recommendations and any other algorithm or classification rule are really useful.
Another important point is related to the aetiology of the cACLD. The Baveno VI guidelines for ruling in CSPH are specifically directed at viral related patients. The reason for that is that the amount of scientific evidence available for other aetiologies, especially for patients with non-alcoholic steatohepatitis (NASH), is low. In that sense, it is very important to include patients with aetiologies other than viral related in the present study. However, at the same time, mixing patients with alcoholic steatohepatitis and those with NASH might considerably mask the results and conclusions, especially considering that alcoholic patients tend to have higher liver stiffness compared with any other aetiology. It is then very important to first perform studies such as the present one by aetiologies and then add up the different aetiologies if the classification rules are equal.
A third point would be whether it is more important to rule in or rule out CSPH in patients with cACLD. Ideally, a perfect test of classification rule should be almost perfect for both ruling in or out. This is rarely the case for any classification rule, but for CSPH, it seems plausible that ruling in is more clinically important and might be even more important if therapies directed at preventing hepatic decompensation become available. In the present study, the proposed L-SWE cut-off (≥16 kPa) for ruling in CSPH seems a little low (86% positive predictive value) and probably this should be raised to achieve at least 90%.
Finally, although using a sequential combined strategy with L-SWE and S-SWE is a nice way of classifying CSPH, it rule might be simplified further for practical purposes. Looking at Jansen et al's data, it seems that using only S-SWE for both ruling in and out CSPH might be sufficient, avoiding the need for handling two parameters and different cut-offs.
Contributors The three authors made substantial contributions to the conception or design of the work and interpretation of data. JG made the drafting of the work; SA and MP revised it critically for important intellectual content. The published version of the letter was approved by all three authors.
Funding SA is a recipient of a JR grant from Instituto de Salud Carlos III, Spain. MP is a recipient of a PFIS grant from Instituto de Salud Carlos III. JG is a recipient of a Research Intensification grant from Instituto de Salud Carlos III. MP is a PhD student at Universitat Autònoma de Barcelona, Spain. The study was partially funded by grants PI12/01759 and PI15/00066 from Instituto de Salud Carlos III, Spain, and cofinanced by the European Regional Development Fund (FEDER). CIBERehd is supported by Instituto de Salud Carlos III, Spain.
Competing interests None declared.
Provenance and peer review Not commissioned; internally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.