Evaluation of Non-invasive Scores for Predicting Portal Hypertensive Gastropathy Severity in Egyptian Patients with Liver Cirrhosis

INTRODUCTION

Liver cirrhosis is the clinical outcome of numerous chronic hepatic disorders. Portal hypertension (PH), which is portal pressure greater than 10-12 mmHg, is a major consequence of cirrhosis. As a sequel of PH, portosystemic collaterals, esophageal varices (EVs), and portal hypertensive gastropathy (PHG) develop [1,2].

PHG is a pain-free state of ectasia of stomach mucosa and compromised mucosal defense in people with portal hypertension (PH) [3]. Endoscopically, PHG manifests as “snakeskin-like” gastric mucosa and primarily entails gastric corpus and fundus [4]. As per the Baveno II criteria, there are two levels of PHG: mild and severe. Red marks denote severe PHG, whereas mild PHG looks like a mosaic picture with a pale areola. Besides, gastric antral vascular ectasia (GAVE), where bands of erythematous mucosa are interrupted by healthy mucosa (watermelon stomach), almost always affects the antrum [5,6].

There is a wide diversity of reported PHG prevalence from 35% to 80% in cirrhotic patients. PHG is implicated in 2 to 12% of acute upper gastrointestinal bleeding (UGIB), with severe PHG being responsible for about 95% of bleeding PHG; some cases can be potentially fatal. PHG also can present as chronic UGIB in 3 to 26% of cases, resulting in iron deficiency anemia. Nearly 12.5% of PHG hemorrhages result in death [6,7].Esophagogastroduodenoscopy (EGD) is by far the best diagnostic modality of PHG.Endoscopy is an incursive modality with a financial and psychological burden on patients, hence the necessity for non-invasive diagnostic methods establishing optimum selection of EGD endoscopy[8].

Transient elastography (TE; Fibroscan) is a non-invading, ultrasound-based tool that evaluates liver stiffness measurement (LSM). TE offers strong sensitivity and specificity for fibrosis and cirrhosis, and it has gained popularity in recent years [9].

A good relationship between LSM and the existence of PH and EV has been proposed, implying the usefulness of LSM for predicting the occurrence of large EVs. Furthermore, the liver stiffness/ spleen size/ platelet ratio score (LSPS) was found to be a good predictor of EVs [10–12].Therefore, in this study, our goal was to anticipate PHG existence and severity non-invasively by measuring LSM and LSPS versus other noninvasive measures such as platelet count/spleen diameter ratio (PSR) and right lobe diameter/albumin ratio (RLAR) in patients with cirrhosis from Egypt.

METHODS

192 adult cirrhotic patients attending the endoscopy unit of the Tropical Medicine Department, at Tanta University Hospitals, participated in our cross-sectional investigation between June 2019 and June 2022. The study protocol received approval by the Tanta University Faculty of Medicine Research Ethics Committee in April 2019 (code 33090/04/19). Written consent had been taken from all participating individuals.

The sample size was determined using the study of Somboon P. and Hongthanakorn C. in 2015, who stated that LSM mean ± SD was 38.3 ± 18.8 and 30.27 ± 21.78 kPa in PHG sufferers and those without PHG, respectively. Considering an 80% power and a 2-sided significance threshold of 0.05. An estimated 192 cirrhotic patients (96 PHG patients and 96 non-PHG patients) made up the sample size [13].

Four groups of patients were generated: Group 1 (48 patients had both PHG and EVs), Group 2 (48 patients had EVs), Group 3 (48 patients had PHG), and Group 4 (48 patients had neither PHG nor EVs as a control).

The criteria of exclusion were the presence of severe ascites, Hepatocellular carcinoma, or a narrow intercostal space that disables TE examination, severe cardiopulmonary diseases, renal failure. Patients treated with non-selective beta-blockers which improved the PHG severity and progression. Patients having varices that had previously been eradicated. Pre-treatment with a portosystemic shunt. Splenectomy or the inability to measure spleen diameter.

A comprehensive history taking, clinical assessment, laboratory workup (complete blood count, liver functions panel, renal functions), Child-Pugh score, conventional ultrasound, upper endoscopy, and LSM using Fibroscan echosens 502 were done on all the patients.

LSM was performed in the right hepatic lobe by situating the Fibroscan probe on an intercostal space at the point where a transverse line from the xiphoid process crosses with the right mid-axillary line (a minimum of 1 cm under the liver capsule), while the patient is resting flat, completely abducting his/her right arm. A valid 10 measurements with a minimum of 60% success rate and ≤30% inter-quartile range (IQR) were necessary to obtain a valid LSM in kilopascals (kPa). The examination was performed initially using the M probe. Whenever the M probe was unable to obtain measurements, the XL probe was used [14].

EVs were classified using the Baveno classification for grading EVs into small, medium, and large. The Baveno II classification for severity of PHG was used to categorize PHG depending on the following categories: Mucosal pattern: mild findings are given a score of 1, and severe findings are given a score of 2. Red marks: isolated lesions receive a score of 1, while confluent red marks receive a score of 2. GAVE receives a score of 2 if present and 0 if absent. A total score of three or lower indicates mild PHG; a score of four or higher indicates severe PHG [5,15,16].

Statistics

IBM SPSS version 22 for Microsoft Windows (Armonk, NY) was utilized to analyze the data. Analysis of variance (ANOVA) tests with subsequent Tukey post hoc tests were utilized for mean comparisons. The Chi-square test was employed for categorical values. To find correlations among various variables, Spearman's rank correlation coefficient test was utilized. The area under the curve (AUC) was obtained using receiver operating characteristic (ROC) curves. For PHG-related variables, the most discriminating "cut-offs" were selected using the highest Youden Index. Statistical significance was interpreted as P values < 0.05.

RESULTS

Of our 192 cirrhotic patients, 102 (53.1%) patients were males, while 90 (46.9%) were females. Our patients were 56.57 ± 8.38 years on average, and the four groups' variations were statistically non-significant in terms of age or sex. Regarding cirrhosis etiology, HCV came first with 175 (91.1%) of the patients, then Non-alcoholic fatty liver disease (NAFLD), unidentified etiology, Wilson's disease, and HBV in 6 (3.1%), 6 (3.1%), 3 (1.6%), and 2 (1.04%) patients, respectively. (Table 1)

The clinical presence of mild ascites demonstrated a significant difference (P = 0.008), while jaundice and lower limb edema showed insignificant variations among the studied groups. Of our 192 patients, 132 (68.8%) fall into Child-Pugh class A, 52 (27.1%) in class B, and 8 (4.2%) in class C. The four groups differed significantly (P = 0.004). (Table 1)

Complete blood count showed that white blood cell counts in the four groups failed to vary significantly (P = 0.146). Group 2 had a considerably lower hemoglobin (Hb) level than Group 4 (P = 0.004). The platelet counts of groups 1, 2, and 3 were substantially reduced compared to those of group 4 (P < 0.001).  (Table 1)

Serum bilirubin, ALT, and AST were insignificantly different amongst the groups (P > 0.05). Compared to group 4, serum albumin in the other 3 groups was noticeably reduced (P <0.001) and group 1 had significantly lower serum albumin than groups 2 and 3 (P =0.022 and P =0.023). In comparison to group 4, INR was substantially lower in groups 1 and 2 (P = 0.027 and P =0.003). (Table 1)

Between the studied groups, there were notable differences in serum creatinine and fasting blood sugar (P =0.005 and P =0.038). (Table 1)

Abdominal ultrasonic examination did not significantly vary between groups considering right lobe diameter while spleen diameter was considerably enlarged in group 1 than in group 4 (P =0.006). The sonographic presence of non-severe ascites was notably different amongst the studied groups (P =0.031). (Table 2)

In contrast to group 4, the PSR in the other 3 groups was significantly decreased (P <0.001). In comparison to group 4, RLAR was considerably higher in the other 3 groups (P <0.001). Also, in contrast to groups 2 and 3, RLAR in group 1 was considerably higher (P <0.001). (Table 2)

The Fibroscan findings demonstrated that the LSM median was significantly elevated in groups 1 and 2 than in groups 3 and group 4 (P <0.001). The controlled attenuation parameter (CAP) median demonstrated no discernable variances across the four groups (P =0.151). The LSPS was notably elevated in group 1 compared to groups 2, 3, and 4 and elevated in group 2 compared to group 4 (P <0.001). (Table 2)

The endoscopic findings demonstrated that 98 (50%) of our patients had EVs; 36 (37.5%) patients had small EVs, 37 (38.5%) patients had medium-sized varices, and 23 (24%) patients had large EVs. PHG was found in 96 (50%) patients; 66 (68.75%) patients experienced mild PHG, and 30 (31.25%) patients experienced severe PHG. (Table 2)

The Hb level, WBCs, and PHG presence and severity did not correlate significantly (P > 0.05). The platelet count was significantly inversely correlated with PHG existence and PHG severity represented by the PHG score (P <0.001) (Table 3).  Insignificant correlations were detected between serum bilirubin, ALT, AST, or INR and PHG. Serum albumin demonstrated a significant indirect correlation with PHG presence and severity (P <0.001) (Table 3).

Child-Pugh class was directly associated with the existence of PHG (P =0.049). Ultrasonographic findings such as right liver lobe diameter, spleen diameter, and presence of mild ascites showed no significant correlations to PHG presence (P > 0.05). (Table 3)

PSR demonstrated a significant direct correlation with the existence and gravity of PHG (P <0.001) (Table 3). A cut-off level of 851 predicted the presence of PHG with AUC = 0.688, sensitivity 62.5%, specificity 70.8%, PPV 65.4%, and NPV 68.2%. (Table 4) (Figure 1)

A significant direct correlation between RLAR and the existence and severity of PHG was discovered (P <0.001) (Table 3). A cut-off level of > 3.11 had 79.2% sensitivity, 39.6% specificity, 56.7% PPV, 65.5% NPV, and an AUC of 0.648 for the prediction of PHG. (Table 4) (Figure 2)

LSM was directly associated with the existence (P =0.003) and severity (P <0.001) of PHG (Table 3). A cut-off level of > 18.35 kPa predicted PHG presence with an AUC of 0.622, sensitivity of 76%, specificity of 52.1%, PPV of 61.3%, and NPV of 68.5%. (Table 4) (Figure 3)

The CAP median demonstrated no significant association with PHG presence or severity (P > 0.05). LSPS demonstrated a significant direct association with the existence and extent of PHG (P <0.001) (Table 3). A cut-off level of > 2.12 predicted the presence of PHG with an AUC of 0.680, sensitivity of 76%, specificity of 53.1%, PPV of 61.9%, and NPV of 68.9%. (Table 4) (Figure 4)

 Table (1). Patients baseline and laboratory characteristics