Abstract
Background and Aims. Endoscopic variceal ligation (EVL) of esophageal varices alters the portal pressure. We observed the changes in 2D-shear wave elastography (2D-SWE) measurements of spleen and liver following EVL and tried to identify the predictors for rebleeding and mortality at 6 months. Methods. A prospective observational study of 202 patients who underwent EVL for bleeding esophageal varices was done. 2D-SWE measurements of liver stiffness (LS) and spleen stiffness (SS) and spleen volume (SV) were measured half an hour before, 1 hour, 2 weeks, and 6 weeks after EVL. All were followed up for 6 months for rebleeding and all-cause mortality. Results. 83 patients were in child C (41%). Difference in SV, SS, and LS at 2 and 6 weeks from baseline was noted as Delta 2 (2nd week post-EVL - pre-EVL SV, LS, and SS) and Delta 3 (6th week post EVL - pre - EVL SV, LS and SS), respectively. Mean Delta 2 VOL and Delta 3 VOL were lower in the bleeding and mortality groups. Delta 2 SS, Delta 3 SS, Delta 2 LS, and Delta 3 LS were higher in the rebleeding and mortality groups. These changes were statistically significant. AUROC in predicting rebleeding was the highest for Delta 2 VOL (0.773) and Delta 3 LS (0.764) amongst the USG parameters that performed better than MELD score (0.677). AUROC in predicting mortality was the highest for Delta 3 VOL and Delta 2 VOL-0.873 and 0.842, respectively, and higher than MELD’s (0.641). Statistically significant variables in binary logistic regression analysis for rebleeding were Delta 3 LS and Delta 3 SS and none for mortality. Conclusion. LS, SS, and SV change after EVL. Changes in liver and spleen stiffness at 6 weeks from baseline had good diagnostic accuracy for predicting rebleeding at 6 months.
1. Introduction
The development of portal hypertension (PHT) in liver cirrhosis is the key factor in increasing mortality and complications. Hepatic venous pressure gradient (HVPG) is the gold standard in diagnosing PHT, and a value more than 10 mm Hg defines clinically significant portal hypertension (CSPH). Since HVPG measurement is scarcely available and invasive, several noninvasive tests are used as surrogate markers of CSPH. Amongst them, elastography techniques measuring liver stiffness (LS) and spleen stiffness (SS) are the extensively studied ones. Though vibration-controlled transient elastography (VCTE) technique is the most commonly available and the most extensively studied method, it needs a special instrument, and hence elastography machines that can be attached to conventional ultrasound (USG) machines are gaining popularity. Amongst them, 2D-shear wave elastography (2D-SWE) is the most recent one, and it assesses stiffness and related parameters by tracking shear waves propagated through a media. Faster imaging techniques and multiple ROI (region of interest) placements when compared to VCTE make them more attractive [1]. But because of its restricted availability, fewer studies on diagnostic performance are available.
Though elastography techniques are routinely used to screen for high-risk varices in compensated cirrhosis, their role in identifying resolution/persistent varices after a variceal hemorrhage has not yet been studied. Also, endoscopic variceal ligation (EVL), the principal endoscopic modality for bleeding esophageal varices, has been known to alter portal pressure. Hence, we planned this study to investigate whether any changes are happening in the 2D-SWE properties of the liver and spleen after EVL and if changes are happening, whether they are associated with rebleeding and mortality. These changes if identified can help us in better prognostication of patients after a variceal hemorrhage and may even obviate the repeat endoscopy in less risky patients.
Since the survival of patients has improved beyond 6 months in many patients with variceal hemorrhage, we also wanted to explore the percentage of rebleeding and/or mortality at a period of 6 months and its predictors.
2. Materials and Methods
This is a prospective observational study conducted at a university hospital in Kerala, South India, after obtaining approval from the Institutional Ethics Committee (HEC No. 04/27/2019/MCT). 202 patients aged above 18 years, with cirrhosis (diagnosed by standard clinical, radiological, or by histopathological changes) and bleeding esophageal varices undergoing EVL, were taken up for the study. All participants gave written informed consent. We excluded patients with extrahepatic portal vein obstruction, portal/splenic vein thrombosis, mass lesions in the spleen, and who had undergone transjugular intrahepatic portosystemic shunting procedures (TIPSS) or endoscopic sclerotherapy in the preceding 6 weeks.
2.1. Study Protocol
Patients who presented with variceal bleeding and endoscopy showing grade 2 esophageal varices with red color signs or higher grades underwent ligation with multiband ligators. They were treated with vasopressors before and after EVL, followed by an oral nonselective beta-blocker (NSBB) according to the standard treatment guidelines. Patients were followed up, and EVL was repeated every 2 weeks till variceal eradication. Varices were classified as follows: (i)Grade 1: small, straight varices(ii)Grade 2: enlarged, tortuous varices that occupy <1/3 of the lumen(iii)Grade 3: large coil-shaped varices that occupy >1/3 of the lumen
2.2. Shear Wave Elastography Measurements
These patients underwent 2D-SWE measurements using Aixplorer (Supersonic Imagine, France) with a convex probe of 1-6 MHz. 2D-SWE values, expressed in kilopascal (kPa), were mapped with a color-coded two-dimensional image with simultaneous conventional B-mode images. All patients were fasting overnight, as required for their upper GI endoscopy. Two experienced ultrasonologists trained in 2D-SWE image acquisition, who had performed at least 200 2D-SWE measurements obtained measurements. Liver stiffness (LS) was obtained at least 1 cm deep to the liver capsule but less than 6 cm deep from the skin surface. Each LS acquisition consisted of 5 sequential measurements obtained in the upper right hepatic lobe via an intercostal approach with breath held at end-expiration and the patient in a supine position. A 10 mm region of interest was chosen, and size changed if necessary, according to the amount of measurable parenchyma and the locations of large vessels, avoiding blood vessels or portal tracts. The mean value was used. After LS measurement, spleen was visualized in the right lateral decubitus position with the left arm at maximum abduction, and the depth of breathing adjusted to increase the visibility of the spleen and its size measured. Spleen stiffness (SS) and LS were measured similarly. A stiffness measurement was considered valid when it has a . The measurement was considered inconclusive when the spleen/liver parenchyma did not provide a properly color-coded elastography or a valid stiffness value was not obtained.
Splenic volume (SV) was calculated using the standard prolate ellipsoid formula () (Figure 1).
The readings obtained were as follows: (1)Pre-EVL (30 min before EVL) (SV0, SS0, LS0) (SV: splenic volume; SS: spleen stiffness; LS: liver stiffness)(2)Post-EVL (1 hour after EVL) (SV1, SS1, LS1)(3)First relook (at 2 weeks from EVL) (SV2, SS2, LS2)(4)6th week (at 6 weeks from EVL) (SV3, SS3, LS3)
The changes in each value during serial measurements were denoted as Delta where (1)(2)(3)
All these patients will be followed up with a telephonic conversation at 6 months regarding the rebleeding episodes and/ or all-cause mortality. Throughout the article, the following references are used: (1)(2)(3)(4)(5)(6)(7)(8)(9)
2.3. Ethical Concerns
Upper GI endoscopy with EVL for treatment of esophageal varices was performed on all subjects as per AASLD guidelines on the treatment of portal hypertension and variceal hemorrhage. Ultrasound abdomen and shear wave elastography, which are noninvasive procedures, were carried out as part of the study.
2.4. Statistical Analysis
The statistical analysis was performed using IBM SPSS version 25. Continuous variables were reported as mean and standard deviation (SD). Categorical variables were reported as frequency and percentage. Comparison between groups was performed using the independent -test for continuous variables with normal distribution and the Chi-square test or Fisher’s exact test for categorical variables. The paired -test was used to assess the significance of changes in different parameters in different subgroups following EVL. Baseline variables were analyzed, and univariate analysis was performed to find out factors predicting rebleeding and mortality at 6 months. Variables found to have on univariate analysis were evaluated with binary logistic regression analysis.
3. Results
202 patients who underwent EVL as secondary prophylaxis were enrolled, and the CONSORT diagram is shown in Figure 2. The demographics and baseline parameters are given in table 1.
The commonest etiology of liver cirrhosis in males was alcohol (52.6%) and in females, it was NASH (58.6%).47 (23.3%) patients had a rebleeding episode, and 41 (20.2%) patients had died within 6 months. The various parameters of patients were compared between patients those who had a rebleeding episode and those without and those who died and those who survived by univariate analysis. The results are tabulated in Tables 2 and 3.
The Receiver Operating Characteristic ROC was plotted for MELD, and the USG parameters found to have significant association in univariate analysis. With regard to rebleeding, the Area Under Receiver Operating Characteristics (AUROC) of Delta 2 VOL, Delta 2 LS, Delta 3 LS, and Delta 3 SS was higher than that of MELD’s. With respect to mortality within 6 months, AUROC of all the USG parameters found to be significantly associated with mortality at 6 months in univariate analysis had a higher AUROC than that of MELD’s. The AUROC values and the optimum cut-offs of various parameters and their sensitivity and specificity values are shown in Table 4. The AUROC curves of USG parameters and MELD are given in Figures 3–6.
The variables identified in univariate analysis to have a significant association with rebleeding and mortality were subjected to binary logistic regression analysis, and the results are tabulated in tables 5.
4. Discussion
Acute esophageal variceal bleeding is a common but severe, life-threatening complication in patients with cirrhosis and portal hypertension. EVL has largely reduced mortality due to variceal bleeding. Guidelines recommend repeating endoscopies in patients who had their varices tackled by EVL till the varices are demonstrated to be obliterated since there are no noninvasive methods to reliably predict the obliteration of varices [2, 3]. This adds to the burden on both the patients and the infrastructure. Hence, there is a need for a simple, reproducible, and noninvasive predictor of recurrent varices in follow-up endoscopies.
It has been shown that EVL alters the portal pressure by Lo et al. They demonstrated that the majority of patients experienced elevated portal pressure after EVL. The patients who had a reduction in portal pressure after EVL had other major collaterals apart from esophageal varices, compared to patients with elevated portal pressure. This elevation of portal pressure after EVL ligation and lack of collateral circulation was postulated to be an important factor in variceal rebleeding [4].
Recent studies demonstrated that SS is a good predictor of PHT since it reflects both the hepatic and extrahepatic components of PHT. reliably predicts high-risk esophageal varices according to the BAVENO VI consensus [5]. No studies are available that measure the change in SS happening as a result of changes in portal pressure induced by EVL, and whether these changes can be related to rebleeding and mortality at 6 months. Hence, we planned this study to measure the same.
Our study involved 202 patients. 114 (56.4%), 88 (43.6%), and 20 (9.9%) required one, two, and three sessions of EVL, respectively. This was similar to the observation by Ahmed et al. wherein 25 (36.2%) patients had variceal obliteration after the first EVL, while 32 (46.4%) and 12 (17.4%) required two and three sessions, respectively [6].
In our study, 47 (23.3%) patients had a rebleeding episode at 6 months. Though studies on the rebleeding frequency at 6 months following EVL are not available, this can be extrapolated from the rebleeding frequencies of 40.5% at 1 year in the study by He et al and 19% in the median follow-up of months in the study by Kumar et al. [7, 8].
Of the 47 patients who had rebleed, 14, 12, and 21 (29, 16, and 25%) were in child A, B, and C groups, respectively, . This could be due to the confounding effect of HCC, as the proportion of patients who had rebleed was 10 : 11 : 16 (27, 30, and 44%) in child groups A, B, and C, respectively, when HCC was excluded.
The endoscopy findings that had a significant association with rebleeding were the presence of portal hypertensive gastropathy (PHG) and gastroesophageal varices (GOV) at first relook scopy. PHG was mild in 30 (20%) and severe in 17 (34%) patients. GOV 1 and GOV 2 were found in 18 (46%) and 2 (13%) patients who bled again. This can be explained based on the fact that PHG has been significantly associated with the severity of portal hypertension as noted by Kumar et al and Kim et al [9, 10].
On univariate analysis, MELD was found to be higher in patients who have had a rebleeding episode ( vs. , ). This is in concurrence with the observations by Chen et al. and Wang et al who had validated the utility of MELD scores in predicting rebleeding at 6 weeks [11, 12].
In this study, 41 (20.2%) patients had died by the end of 6 months which can be again extrapolated from the mortality rates of 40% observed at 1 year by Sharma et al. [13].
On univariate analysis, child distribution (A:B:C) (8 : 7 : 26), (17 : 9 : 31%, ) and higher mean MELD scores ( vs. , ) were found to be significantly associated with mortality. This is in concurrence with the previous observation by Chen et al that MELD is a good predictor of mortality after variceal bleeding at 6 weeks [11, 14, 15].
In our study, mean Delta 2 VOL and mean Delta 3 VOL were significantly lower in patients who had rebleeding than in those who did not. The values of mean Delta 2 VOL and Delta 3 VOL were vs. cm3, , and vs. cm3, , respectively. Similarly, mean Delta 2 VOL and mean Delta 3 VOL were lower in the mortality group ( vs. cm3 and vs. cm3, respectively) than in the survivors.
Mean Delta 2 SS and Delta 3 SS were significantly higher in the rebleeding group ( vs. , ) ( vs. , ), respectively. Similarly, Delta 2 SS and Delta 3 SS were higher in the group with mortality than in the group that remained alive at 6 months. The values were vs. , , and ( vs. , ), respectively.
This reduction in SV with an increase in SS in the rebleeding and mortality group can be hypothesized to the lack of collaterals and worsening of PHT. This impact of collateral circulation has been shown by Tarantino et al that the median splenic vein flow velocity in patients with splenorenal shunts was significantly inferior to that of patients without them [16]. This reduction in splenic vein flow velocity could be the explanation for the decrease in SS in patients with significant extra esophageal collaterals and consequently no rebleeding episodes at 6 months. They are probably having less severe PHT and fewer complications.
Another observation in our study was that Delta 2 LS and Delta 3 LS were significantly higher in the rebleeding group than in the other vs. , ) and ( vs. , ), respectively. Also, Delta 2 LS and Delta 3 LS were significantly higher in the mortality group than in the survivors. The values were vs. , , and vs. , respectively.
This can be explained based on the observation by Piecha et al who evaluated the changes in HVPG with EVL and TIPS. They noted that LS increased transiently in the majority following EVL and in few after TIPS. They postulated that the collateral circulation determines the changes in stiffness happening after EVL/TIPS. Patients with more fibrosis/severe disease can have an increase in stiffness while those with lesser fibrosis and/or more collaterals had a reduction in stiffness [17]. Those who had rebleeding/mortality had probably lesser collaterals and consequent worsening of PHT.
The variables identified to be significant in univariate analysis were subjected to binary logistic regression analysis. Delta 3 SS and Delta 3 LS were found to have a significant positive correlation with rebleeding with an odds ratio (OR) of 1.075, 95% CI (1.03-1.12), and 1.156 95% CI (1.08-1.23), respectively. None of the variables were predictive of mortality.
The AUROC of MELD for the prediction of the rebleeding episode and mortality was 0.677 and 0.641, respectively. With a , the sensitivity and specificity were 51% and 76%, respectively, for a rebleeding episode and 48% and 75%, respectively, for predicting mortality.
The AUROC of Delta 2 VOL, Delta 2 SS, and Delta 3 SS for predicting rebleeding was 0.773, 0.674, and 0.706, respectively, and higher than that of MELD. With a cut-off of ≤46.9 cm3, the sensitivity and specificity of Delta 2 VOL were 76% and 72%, respectively. The cut-offs for Delta 2 SS and Delta 3 SS were >9.8 and >21 kPa, respectively. The sensitivity and specificities were 87% and 43% for Delta 2 SS and 63% and 72% for Delta 3 SS, respectively.
The AUROC of all the ultrasound parameters in predicting mortality was more than that of MELD’s. The highest among them were for Delta 2 VOL (0.842) and Delta 2 LS (0.841). At a cut-off of ≤44 cm3, the sensitivity and specificity of Delta 2 volume were 85% and 74%, respectively. The cut-off of Delta 2 LS was >11.9 kPa, and sensitivity and specificity were 63% and 93%, respectively.
Though there are studies regarding the utility of 2D-SWE in diagnosing CSPH, no studies are available regarding the change in the elastography parameters that happen as a result of EVL-induced changes in portal pressure. Our study shows that significant changes are happening after EVL, and they can be used to predict rebleeding and mortality. Especially, the difference in LS and SS at 6 weeks from the baseline is found to be statistically significant in binary logistic regression analysis to predict rebleeding at 6 months. Our study needs further validation in a larger cohort.
We admit that the study has limitations. The exact reason for the varied response in portal pressure could not be ascertained. The presence of significant portosystemic collaterals was hypothesized to be the reason but it was not objectively demonstrated. But to the best of our knowledge, this is the first study of its kind to look at changes happening in elastography properties of the liver and spleen and its usefulness in predicting rebleeding and mortality at 6 months in patients undergoing secondary prophylaxis for esophageal varices.
If validated in larger cohorts, these noninvasive tests can identify people at increased risk of recurrent variceal hemorrhage and mortality enabling increased surveillance in this high-risk population.
5. Conclusion
Liver and spleen stiffness as measured by 2D-SWE and spleen volume changes after endoscopic variceal ligation and is reflective of the dynamic nature of portal hypertension. The difference in spleen volume from baseline at 2 weeks and 6 weeks after EVL is significantly lower in both the rebleeding and mortality group than in the nonrebleeders and survivors. Also, the difference in spleen and liver stiffness at 2 weeks and 6 weeks after EVL is higher in the rebleeding group and those who had died by the end of 6 months than in the nonrebleeding group and group who survived. The difference in liver and spleen stiffness at 6 weeks had a good diagnostic accuracy for rebleeding at 6 months.
Data Availability
The data used to support the findings of this study are included within the manuscript.
Disclosure
This research paper was presented in the Indian National Association for the Study of the Liver-28th Annual scientific Meeting held virtually from 6th to 8th August, 2021, in Young Investigator Award session.
Conflicts of Interest
The authors declare that they have no conflicts of interest.