Possible use of 2D shear wave liver elastography in new-onset ascites evaluation

Ascites means the appearance of free fluid in the abdominal cavity. The most common processes involved in its occurrence are portal hypertension (PH) and peritoneal membrane disease. In the Western population, the most common cause of ascites is PH, which is due to cirrhosis in most cases [1‐3]. The clinically relevant ascites division is based on the serum-ascites albumin gradient (SAAG) determination, which biochemically divides ascites into exudate and transudate. While albumin-rich ascites (exudate) is due to inflammatory and/or neoplastic causes, ascites due to PH is a transudate characterized by SAAG values ≥ 11 g/L; this value has more than 97% accuracy in differentiating portal hypertensive causes from other causes [1‐3]. However, several etiologies may be present simultaneously (mixed ascites), which complicates differential diagnosis [2, 4‐6]. The most common causes of ascites are portal hypertension, neoplastic causes and inflammatory causes, or pancreatic ascites.

In particular, it is clinically very important to rapidly classify ascites into benign (mostly due to PH) and malignant (due to peritoneal carcinomatosis); more than 90% of new-onset ascites cases belong to one of these two etiologies [2, 6, 7]. In the presence of ascites, findings typical of cirrhosis such as morphological changes and PH-specific signs such as the presence of portosystemic collaterals (Table 2) can be used to postulate PH as the etiology [8, 9]. Imaging signs of cirrhosis and PH have high inclusion and low exclusion value.

Liver elastography is commonly used to further improve the cirrhosis and PH diagnosis [8, 9]. According to the Baveno VII consensus conference [10], the presence of clinically significant portal hypertension (CSPH) in patients with compensated advanced chronic liver disease (cACLD) can be reliably ruled out by liver elastography when the elastographic value is < 15 kPa and platelet count > 150 × 10⁹ per nanoliter of blood. The confirmatory threshold of elastography for CSPH in the most common cACLD etiologies is > 25 kPa, and from 15 kPa onward, the probability of CSPH presence slowly increases [10]. 2D shear wave elastography (2D-SWE) is a well validated method in this field [11‐13]. In two studies performed in a subset of patients with cirrhosis with or without ascites, 2D-SWE showed a significantly positive correlation with invasively evaluated portal pressure [14, 15].

A study by Kolhaas et al. on a phantom model and using TE (XL probe) found that, in cases where the thickness of the perihepatic ascites layer did not exceed 20 mm, a non-hepatic cause of ascites could be reliably identified [16]. A clinical study performed by Bota et al. identified a good predictive value of point shear elastography (pSWE) to distinguish cirrhotic and non-cirrhotic etiology of ascites [17].

Given the wide availability of elastographic methods, we are interested in whether 2D-SWE can provide data allowing a non-invasive discrimination between PH-related causes and other causes in patients with new onset ascites. There is no study data in this area. Also, the applicability of 2D-SWE measurement using Esaote MyLab Nine apparatus in patients with ascites will be concurrently assessed.

We obtained the approval of the regional General Hospital Celje Ethics Commission and performed a retrospective analysis of the prospectively collected data. Informed written consent was obtained from each patient included in the study. The study protocol conformed to the ethical guidelines of the Declaration of Helsinki and Istanbul.

The main characteristics of the study population are provided in Table 3.

Our study included 34 patients (65%) with new-onset ascites with SAAG ≥11 g/L (Group 1). In these cases, cirrhosis and accompanying PH has been confirmed. Etiology of cirrhosis was due to harmful alcohol consumption (20 patients) and metabolically-induced steatohepatitis (8 patients) or a combination of both conditions. Cirrhosis was due to chronic HCV infection and primary biliary cholangitis in two patients.

Group 2 included 18 patients (35%) with new-onset ascites with SAAG < 11 g/L; ascites was due to cancer in 13 patients (25%) and in all cases with cytologically confirmed peritoneal carcinomatosis. Causes of peritoneal carcinomatosis included: upper and lower gastrointestinal cancer (four patients), ovarian cancer (three patients), pancreatic cancer (three patients), breast cancer (two patients), and primary peritoneal cancer (one patient). The remaining five patients (10%) had pancreatogenic ascites (two patients), tuberculosis (one patient), obscure cause (one patient), and immune-mediated ascites (one patient). To evaluate the presence of sinusoidal PH, three patients in Group 2 (5% of the total cohort) underwent an HVPG assessment. A total of 27 patients (52%) were deceased by the end of the study.

The main finding of this study at a single secondary hospital center is that 2D-SWE of the liver can noninvasively discriminate between portal hypertension-related newly onset ascites. The proposed 14 kPa cut-off value is comparable to the values proposed for cirrhosis confirmation in the cACLD group [10, 18‐20] and roughly comparable to the proposed values for noninvasive CSPH exclusion criteria in published studies and meta-analyses (cACLD populations) [6, 10‐15, 24]. The result has important clinical value since our study defines liver elastography (noninvasive, bedside, widely available method), especially its 2D-SWE version, as the best investigative method for assesing CSPH in patients with newly formed ascites. This definition applies mainly to patients with dACLD in the stable disease phase and may be important for further study in patients with severe CSPH (HVPG > 16 mmHg). According to our analysis, the close relationship between 2D-SWE and CSPH is also indirectly indicated by the characteristic association between SAAG, 2D-SWE, and platelet blood count. Thrombocytopenia is one of the characteristic CSPH consequences and is involved in most of the proposed noninvasive CSPH assessments guidelines [10].

Similar conclusions were reached by Bota et al. in a larger sample of patients where the liver fibrosis assessment was performed using pSWE [17]. Due to a different design (strictly sequential patients inclusion), their study included significantly more patients with cirrhosis and a comparable number of patients in the exudative ascites group. The cut-off values ​​of their study and ours differ (approximately 10 kPa in Botaet al. versus 14 kPa in our cohort). An important reason for this difference may be the used elastographic method, as 2D measurement performs liver stiffness analysis on a much larger liver parenchyma sample. Furthermore, the Bota et al. cohort does not clearly state whether the cut-off value analysis was performed under the SAAG < or ≥ 11 g/L condition. Furthermore, patients with cirrhosis and the rest of the group ratios are significantly different (approximately 3:1 in Bota et al. versus 2:1 in our study). The accuracy of the proposed cut-off value between their cohort and ours was otherwise comparable (95% accuracy versus 94%).

Our study’s calculated 14 kPa cut-off value had an appropriate exclusion value for identifying patients with ascites not due to PH, especially in the patient group with peritoneal carcinomatosis resulting from metastatic cancer. Together with some other study data in this area [25‐34] – elastography can provide additional non invasive diagnostic information.

Although the analysis was performed in a minimal number of patients, the 2D-SWE exclusion value determining the cause of ascites in groups other than liver cirrhosis or peritoneal carcinomatosis appears to be less accurate, especially in cases where etiologic differentiation was difficult and required extensive and invasive diagnostic work-up. A study by Bota et al. met similar findings [17]. Furthermore, 2D-SWE does not provide sufficient information on mixed ascites, where paracentesis is always required for a definitive determination. The same applies to the cytological pattern analysis and inflammatory process evaluation in ascites [35‐41].

Our study used SAAG, a commonly used sum in this field, to divide the ascites cause. A study on several patients by Li et al. used a serum ascites cholesterol gradient (SACG) to determine the cause of ascites. SACG proved to be comparable to SAAG in the PH ascites group and more sensitive in differentiating the cause in the mixed ascites and non-PH ascites group but was not evaluated in our study [42].

Regarding the 2D-SWE measurement using MyLab Nine and its clinical applicability, our study shows about an 85% success rate in performing reliable measurements. This percentage is significantly lower than those reported in studies that evaluated liver cACLD cohorts (typically > 90 or even 95% applicability) [14‐16, 24, 43, 44]. There are several reasons for this finding. In our study, the elastographic measurement was performed exclusively in patients with ascites. According to some studies, ascites should not significantly affect the elastographic result quality [14‐16, 24, 44]. On the other hand, according to Zhang et al., it had a significant effect on both reliability and measurement quality, showing a similar effect on the end result as ROI size [45], with which the authors of our study agree.

It is also important to mention the lack of a protocol for reliable 2D-SWE [22]. The protocol in our study was a sum of study recommendations and past findings in this area [18‐22]. The ultrasound machine type (MyLab Nine) also had a significant impact because there is no official validation yet compared with commonly available devices on the market (Supersonic Imagine, France) [20]. The results between different measurement systems can vary between 6 and 12% [19]. A recently published study provided head to head 2D-SWE comparison between MyLab Nine and Supersonic Imagine in a cACLD population in order to assess liver fibrosis staging tresholds with very good proven correlation [46]. To our knowledge, our study is the first one to evaluate the MyLab Nine 2D-SWE measurement in the dACLD group. The most important causes of unreliable measurement were comparable to those defined to date; however, the influence of the measurements’ depth, especially with concomitant overweight, bilirubin level, patient participation, and the heart function’s influence on liver movement during each measurement, is more pronounced in patients with significant ascites [15, 18, 19, 21, 47‐50].

Our study has important limitations, which we must consider when interpreting the results. Our study was conducted in one hospital center with a relatively small group of patients and retrospective data analysis. Given these findings, 2D shear wave liver elastography has potential in evaluating new-onset ascites but needs further validation in a larger prospective study. Patients in our cohort had strictly defined inclusion and exclusion criteria due to the potential impact on the SAAG and 2D-SWE results. In our estimation, this decision may have excluded up to 30% of the total patients with newly onset ascites. Elastographic measurement was also performed by a single investigator and on an US apparatus that is not study validated, meaning that we do not have data on inter-observer reproducibility and on head to head comparison with the validated US soft-ware. In the group of patients with transudation resulting from PH and cirrhosis, we did not precisely define the liver cirrhosis etiology influence on the ascites formation and elastographic measurement. Recent data suggest that, especially in the group with metabolically-induced liver cirrhosis, ascites in a small group of patients develop in parallel or before the proposed HVPG threshold values [50]. Also, it is important to note, that our study does not provide direct 2D-SWE comparison between dACLD and patients with normal liver stiffness which can explain why no false negatives were detected when performing ROC analysis in the Group 1. As such, it can not be fully translated to the populations with low PH-related ascites prevalence. The presence of free fluid is likely to make the accuracy of elastography suboptimal. Since there is no comparative clinical study and no gold standard to demonstrate the elastography accuracy in patients with ascites, we have to carefully interpret the results even in the most reliable measurements cases. Finally, paracentesis is still a required diagnostic procedure when evaluating new-onset ascites and can probably paradoxically enhance 2D-SWE quality as already mentioned.

The results of our study underline the feasibility, and good applicability of a simple and readily available ultrasound method to assess the PH-related cause of newly onset ascites at the bedside. Also, 2D-SWE held the accuracy necessary for differentiating between transudative ascites in patients with PH related to liver cirrhosis and exudative ascites in patients with underlying peritoneal carcinomatosis. In our view, the routine use of this method during ultrasound examinations of patients with newly onset ascites should be attempted to enable quick, noninvasive identification.