In children, portal hypertension (PHTN) is a consequence of a wide spectrum of cirrhotic and noncirrhotic pediatric liver diseases. Variceal bleed is the most dreaded complication of PHTN. Approximately, 50% of pediatric patients with cirrhosis and 90% of those with extra hepatic portal vein obstruction (EHPVO) will experience variceal bleeding .
Hepatic venous pressure gradient, used to define clinically significant PHTN in adults, is technically challenging in children [2,3]. In children, diagnosis of clinically evident PTHN is based on splenomegaly (>2 cm), thrombocytopenia (<150 000), ascites, or endoscopic evidence of varices .
Though screening endoscopy is recommended in all fresh adult chronic liver disease (CLD) cases, this is not the standard practice in children . As of now, EGD remains the gold standard for assessment of varices. However, esophagogastroduodenoscopy (EGD) being invasive and needing technical expertise, there is an unmet need for developing noninvasive tools (NITs) for screening children with PHTN at high risk for varices and to minimize unnecessary endoscopies in those without varices.
Elastography techniques in pediatrics have been primarily used for the assessment of liver stiffness [6,7]. Liver stiffness measurement (LSM) measures only the intrahepatic component of PHTN, whereas splenic stiffness measurement (SSM) in addition assesses the prehepatic and splanchnic components of PHTN; hence, SSM is hypothesized to be a better predictor of PHTN and presence of varices than LSM.
In this study, we aimed to evaluate the utility of splenic (SSM) and liver stiffness (LSM) by sonoelastography based on principles of point shear wave elastrography (pSWE), as well as blood-based NITs, for predicting varices in children with PHTN of different etiologies, considering endoscopy as the reference standard.
This was a prospective case-control study. A total of 105 newly diagnosed children with PHTN, confirmed by clinical examination (splenomegaly and significant ascites), ultrasonography, liver histology (advanced fibrosis or nodules), or endoscopy, were screened over a period of 18 months. Splenomegaly because of other etiologies (hematological/storage/infectious) and patients already on therapy for PHTN were excluded. Among those screened, 85 cases were enrolled for the study: 4/105 children were uncooperative for elastography; 13 cases defaulted on their appointments or did not give consent and the remaining three fulfilled the exclusion criteria: two had storage disorder and hemolytic disease and a third was already on beta blockers when screened. Ninety-seven age-matched healthy children, with normal growth and sonography were included as controls (flow of study; Fig. 1).
EGD was the reference standard; all 85 enrolled cases underwent EGD to assess variceal status by an experienced pediatric gastroenterologist after sedation with ketamine (1 mg/kg), using standard forward viewing endoscope (Olympus Optical Co. Ltd, Tokyo, Japan) after an overnight fast. PHTN cases with varices were classified as varices present and those without varices as varices absent. Those with varices present were also subclassified into high risk (HV) [based on the presence of grade III varices, grade II with red color sign (RCS)] or low-risk (LV) group (based on the presence of grade II without RCS and all grade I varices) .
Measurement of splenic stiffness
SSM was performed by an experienced radiologist, blinded to the endoscopic findings, within a week of endoscopy. The iU22 ultrasound system (Philips Medical Systems, Bothell, Washington, USA) was used with a 2-to 5-MHz broadband convex array transducer. All cases were examined in supine position. Region of interest of prefixed size was taken at three different sites in the spleen: in upper, mid and lower poles of spleen avoiding major vascular structures, and a mean was calculated. Results were displayed in kilopascals (kPa). A similar protocol was followed for SSM in study controls.
Measurement of liver stiffness
LSM was measured in 81 cases. A total of three readings (values in kPA) from different liver segments were taken using the same principle.
Measurement of blood-based noninvasive tools
The following NITs were assessed:
- 1. Thrombocytopenia: low platelet count is the oldest tool for noninvasive prediction of PHTN .
- 2. Spleen size Z score (SAZ): absolute spleen size (ultrasonographic) cannot be used in children as spleen size varies with age and BMI. So, SAZ was derived as given below. Because of paucity of age-specific data in normal Indian children, we took normative data as suggested by a study by Megremis et al. [9,10].
SAZ = Spleen size observed (cm) – Spleen size of age and sex matched healthy child (cm)/SD score
- 3. Platelet spleen size Z score (PSZ): it was calculated as a ratio of platelet count (109 cell/µL) divided by SAZ as previously described .
- 4. Clinical prediction rule (CPR): based on platelet, SAZ and albumin were calculated as below . CPR = [0.75 × platelet (109/µL)/SAZ + 5] + 2.5 albumin (g/L).
- 5. King’s variceal prediction score (KVaPS): based on albumin and adult spleen size equivalent is calculated as 
KVaPS = 3 albumin (g/L) – 2 Expected adult spleen size (EASS)
EASS = 9.9 + SAZ × 1.27 for females and 11.9 + SAZ × 1.49 for males.
- 6. Aspartate aminotransferase platelet ratio index (APRI): calculated as ratio of observed aspartate aminotransferase (AST) upon upper limit of AST for patient’s age divided by platelet count multiplied by hundred. An APRI cutoff of 0.6 has been suggested as a screening tool for varices, with a sensitivity and negative predictive value of 100% .
Differences between means of continuous nonnormally distributed variables were assessed with the Mann–Whitney rank-sum test. The Chi-square or Fisher’s exact tests was used for comparisons of categorical variables. A receiver operating characteristic (ROC) curve was constructed, and the area under the curve (AUROC) was calculated with the corresponding 95% confidence interval (CI). Statistical analysis was done using the statistical software ‘SPSS version 22’ (SPPS Corp, Chicago, Illinois, USA).
The study was approved by Institutional Ethical Clearance Committee (PGIMER Chandigarh) and is registered under Indian Registry of Clinical Trials (CTRI/2018/07/014929).
A total of 85 cases of PHTN and 97 healthy age-matched controls were enrolled in the study over a period of 18 months. Median age of cases was 7 (3–10) years with 52 (61%) males. The controls had a median age of 7 (3.75–9) years and 67 (69%) were boys. Of 85, 27 (32%) of the cases had bled and 23/85 (27%) had ascites. There were 26 (30.5%) prehepatic, 55 (64.8%) hepatic and 4 (4.7%) posthepatic causes of PHTN. The prehepatic PHTN cases were exclusively due to extrahepatic portal vein thrombosis EHPVO. The remaining 59 (70%) cases of PHTN were grouped as CLD.
The CLD subgroup (Child Class A, B, C: 44.8, 36.2, 19%, respectively) had heterogeneous etiology. Among the 59 children with CLD, 15 (25%) were Wilson’s disease, 10 (16%) autoimmune hepatitis, 7 (11%) post-Kasai biliary atresia, 5 (8%) other cholestatic liver disease (Progressive Familial Intrahepatic Cholestasis and Paucity of Intrahepatic Bile Ducts), 4 (6%) Hepatic venous outflow tract obstruction, 3 (5%) congenital hepatic fibrosis, 5(8.4%) other metabolic liver disease, 2 (3.4%) each Indian childhood cirrhosis and drug-induced liver disease, one Non Alcoholic Steatohepatitis and the remaining 5 (8.5%) were idiopathic.
The EHPVO subgroup was a homogenous and younger cohort, mainly idiopathic, with a median age of 5 years. Variceal bleed was a presentation in 81% of the EHPVO cases, versus 8.4% of the CLD cases. Significant ascites was seen in 3.8% of EHPVO versus 62% of CLD. EHPVO cases had more severe thrombocytopenia with normal transaminase levels compared with the CLD group (median platelet 0.90 versus 1.49/mm3, median alanine aminotransferase of 40 versus 67 P < 0.01, P < 0.01). Sonography revealed larger liver in CLD than in EHPVO (mean liver size 12 versus 9.3 cm, P < 0.001) with a comparable spleen size.
Fifty-five (65%) cases of PHTN had evidence of varices on EGD (varices present) and 30 (35%) cases had no varices (varices absent). High-risk varices (HV) were seen in 32 (58%) cases and 23 (42%) cases had low-risk varices (LV). Baseline comparison between varices absent and present groups are provided in Table 1.
Splenic stiffness measurement/liver stiffness measurement
Success rate of elastography was 97% in our study. We compared SSM and LSM values between cases versus controls; varices present versus varices absent in the whole group and EHPVO and CLD subgroups. The data for all comparisons are shown in Table 2.
SSM and LSM were higher in cases as compared with age-matched healthy controls (SSM cases versus controls: 5.5 versus 3.8 kPa, P < 0.001; LSM cases versus controls: 5.0 versus 2.9 kPa, P < 0.001). The optimal cutoffs of SSM and LSM for discriminating cases from healthy controls were 3.8 and 3.2 kPa with an AUROC of 0.67 and 0.77, respectively. SSM but not LSM was significantly higher in children with varices than those without varices (SSM in varices present versus absent: 6.2 versus 4.1, P value <0.015; LSM in varices present versus absent: 5.7 versus 4.5 kPa, P < 0.197). The AUROC for SSM for variceal prediction in our study cohort was 0.68 (0.56–0.81) at a cutoff of 4.5 kPa. SSM was higher in those with variceal bleed as compared with nonbleeders (SSM bleeder versus nonbleeder: 7.1 versus 5.1 kPa, P < 0.017) with AUROC of 0.66 at a cutoff of 6.0 kPa. SSM correlated positively with spleen size of cases. The larger spleens had higher splenic stiffness (Spearman corelation of spleen size and SSM cases versus controls: 0.357, P < 0.001 and 0.152, P = 0.159). Histological assessment of liver fibrosis was done in 26 cases but failed to show significant correlation with stiffness.
The data for SSM and LSM were also analyzed separately in the CLD and EHPVO subgroups.
Chronic liver disease (subgroup analysis)
Both SSM and LSM were significantly higher in CLD group with varices than those without (SSM varices present versus absent: 6.2 versus 3.8 kPa, P = 0.002; LSM varices present versus absent: 8.2 versus 4.7 kPa, P = 0.005). The AUROC for SSM was 0.73 (0.60–0.86) and for LSM was 0.71 (0.57–0.86) for predicting varices in the CLD subgroup at a cutoff of 5.2 and 6.4 kPa, respectively (Table 2; Fig. 2a).
Extra hepatic portal vein obstruction (subgroup analysis)
Cases of EHPVO with varices had a higher SSM, but LSM failed to show significant difference (SSM varices present versus absent: 29.7 versus 5.8 kPa, P < 0.015; LSM varices present versus absent: 3.8 versus 2.2 kPa, P = 0.138). The AUROC for SSM was 0.94 (0.83–1.0) for predicting varices at a cutoff of 12.4 kPa (Table 2; Fig. 2a).
Blood-based noninvasive tools
The blood test-based NITs (CPR, PSZ, KVaPS and APRI) were performed for 83 PHTN cases; CPR, KVaPS and APRI were able to predict varices in the CLD subgroup. PSZ was not an effective tool in our study. (CPR: varices present versus absent: 73 versus 145, P < 0.001; KvaPS: varices present versus absent: 37.9 versus 101, P value < 0.001; APRI: varices present versus absent: 2.3 versus 1.1, P < 0.001). The NITs had excellent AUROCs for predicting varices in the CLD subgroup. However, they were not found to be useful in the EHPVO subgroup (Table 3; Fig. 2b).
The present study was conducted with the aim of evaluating the utility of NITs including blood markers and liver/splenic stiffness by ultrasonography based on principles of pSWE, as a predictor of variceal status in children with PHTN of different etiologies considering endoscopy as a reference standard. pSWE was chosen over transient elastography, as it does not require a separate machine (program being incorporated), allows real time view of the viscera and choice of region of interest, and moreover, the presence of ascites and obesity do not preclude its use for measuring stiffness.
The SSM and LSM of the healthy controls in our study were 3.8 kPa (2.5–5.9) and 2.9 kPa (2.1–3.6). Spleen and liver stiffness in healthy children have been assessed previously: a study using SWE (ARFI) for healthy children found normal mean shear wave (pSWE) velocity of 1.12 m/s for liver and 2.25 m/s for spleen . Another study using ARFI found SSM to increase with age (healthy infants versus children: 2.01 and 2.32 m/s, P < 0.001) . Normal values of LSM in children in a multicentre study were obtained as 3.3, 4.1 and 4.1 kPa using transient elastography, pSWE and 2D SWE . Adult studies using ElastPQ (pSWE) technique, found LSM of 3.5 kPa and SSM of 16.6 ± 2.5 kPa in healthy adults using SWE [16,17]. The higher values of SSM in adults may be reflective of increasing reticuloendothelial hyperplasia in spleen with increasing age.
In our study group, both SSM and LSM values were significantly higher in cases as compared with the controls. When all etiologies of PHTN were clubbed together, SSM but not LSM was predictive of varices. SSM, in our study, had a comparable sensitivity for prediction of varices, to that of other studies. Most pediatric studies have used transient elastography for elastography. Goldschmidt et al, used transient elastography for SSM and found it to be 75 kPa in PHTN cases with varices and 24 kPa in the nonvariceal group. They had concluded that SSM <22.8 kPa rules out the presence of varices . Another pediatric study using transient elastography found SSM as the best predictor of clinically significant varices with AUROC of 0.92 with an optimal cutoff value of 38 kPa with high sensitivity of 89% and specificity of 82% . Both these studies, however, had included only children with clinically significant PHTN (large spleens and thrombocytopenia), who already had a high pretest probability of having varices and thus, had higher average SSM than is expected in newly diagnosed cases with any degree of PHTN. A Turkish study using pSWE found an AUROC of 0.74 for SSM with cutoff of 2.09 m/s having sensitivity of 80% and specificity of 77% . The lower cutoffs of SSM for variceal prediction in our study are due to the use of a different technique of elastography (transient elastography versus pSWE); this could also be explained by the unbiased enrollment of consecutive freshly diagnosed cases of PHTN, as they presented to us, irrespective of the platelet count/spleen size. It is possible that our cases being freshly diagnosed had less rediculoendothelial hyperplasia and splenic congestion, leading to lower SSM. An early adult meta-analysis found limited role of SSM in predicting PHTN, but a second meta-analysis found SSM a better predictor of varices than LSM, similar to our results [21,22].
There are contradicting results of efficacy of LSM in variceal prediction. The King’s College group and Goldschmidt et al.  did not find LSM to be a predictor of varices. In contrast, another pediatric study found LSM to be a better predictor than SSM using pSWE for PHTN . The adult meta-analysis found an LSM predictive of varices/PHTN; however, it was of lesser value than SSM .
The variation in predictive ability between SSM and LSM was better understood in our subgroup analysis (Table 3). Though both SSM and LSM predicted varices in the CLD subgroup, in the EHPVO subgroup, only SSM was a tool for variceal prediction. This is because of the largely preserved hepatocellular function in children with EHPVO.
Another point to be noted is that the cutoff of SSM for prediction of varices in EHPVO was significantly higher than that for CLD (12.8 versus 5.8 kPa). This is probably explained by the spleens being stiffer in EHPVO; as the PHTN in them is attributable to a prehepatic component, unlike CLD where it is the intrahepatic component of PHTN that plays a major role. EHPVO and CLD are thus two heterogeneous groups which need to be assessed as separate entities and different cutoffs for each need to be derived for prediction of varices.
Another issue to be kept in mind is the higher propensity of children with EHPVO (81%) to bleed as compared to children with CLD. Therefore, the clinical threshold for screening endoscopy in EHPVO is lower as they are more likely to have high-risk varices. However, bleeds in EHPVO are better tolerated than CLD. It is actually more pertinent to be able to predict varices in the latter group to improve outcomes. In this context, our study found blood-based NITs better tools to predict varices than elastography in CLD.
Adult EHPVO cases had a transient elastography SSM cutoff for varices as 42.8 kPa with a sensitivity of 88% and specificity of 94% . Another pediatric study did a subgroup analysis for the EHPVO and CLD groups. They also found higher SSM in the EHPVO group than CLD (62.8 versus 49.6 kPa) and obtained an SSM cutoff of 16.8 kPa for prediction of EHPVO with 100% sensitivity and 100% specificity . An Indian study on adult EHPVO, however, failed to show SSM as an accurate predictor of varices .
Platelet count is the oldest and still the most reliable marker of PHTN used in combination with either elastography or spleen size, or as part of formulas for NITs. In our study, thrombocytopenia by itself suggested an increasing severity of PHTN. The previous study by Gana et al.  had found an AUROC for platelet count of 0.79 at a cutoff of <115 × 109. Platelet:spleen-size Z score ratio (PSZ): it was first used by Gana et al. and confirmed in another pediatric study for predicting varices with a high AUROC of 0.84 [25,26]. However, this ratio failed to show significance in our study. This could be because of the use of western reference data of spleen size for SAZ calculation. APRI: this parameter was first evaluated in adult CLD cases. Its extrapolation to pediatric PHTN cases has been done previously. The AUROC of 0.87 was calculated at a cutoff of 1.92 by Chongsrisawat et al. . In our study, APRI was significant in the overall group as well as the CLD subgroup. CPR: Gana et al.  found an AUROC of 0.80, whereas Adami et al.  found an average AUROC of 0.66. Our study revealed CPR to have the best predictive ability for varices in the CLD subgroup. The CPR was not found as a good tool for EHPVO cases as they tended to have a normal albumin. KVaPS: the initial investigators had found an AUROC of 0.81 with sensitivity and specificity of 86 and 71%, respectively, for KVaPS . Our cases had significant KVaPS in CLD cases. The AUROC was found to be 0.93 with 83% sensitivity and 73% specificity.
Our study is the first to assess the role of SSM using sonoelastography in the pediatric population, from the Indian subcontinent. The cases included were consecutive, unselected newly diagnosed cases of varying severity and etiology of PHTN, all of whom underwent endoscopic assessment. This is reflective of the patient population as it actually presents to the clinician. Our study is thus without selection bias and reflects the real life scenario which pediatric hepatologists would face in their day-to-day practice.
There were certain limitations in our study. SSM failed to show significant difference between high-risk and low-risk varices, though the SSM in bleeders was significantly higher. The cohort of CLD cases who underwent biopsy was small and etiologically heterogeneous and the results failed to show correlation between liver fibrosis and either LSM or SSM. Assessment of SSM using pSWE required a few seconds of breath holding which was difficult and time consuming to implement in young children and infants.
To conclude, SSM by pSWE is a feasible and better option than LSM for predicting varices in children. In the CLD subgroup, both LSM and SSM may predict varices; however, in EHPVO, only SSM is predictive of varices. Different cutoffs are needed to predict varices in the two subgroups. Blood-based NITs outperformed the elastography technique for predicting varices in children in the CLD subgroup.
Conflicts of interest
There are no conflicts of interest.
1. Ling SC, Walters T, McKiernan PJ, Schwarz KB, Garcia-Tsao G, Shneider BL. Primary prophylaxis of variceal hemorrhage in children with portal hypertension: a framework for future research. J Pediatr Gastroenterol Nutr. 2011; 52:254–261
2. Miraglia R, Luca A, Maruzzelli L, Spada M, Riva S, Caruso S, et al. Measurement of hepatic vein pressure gradient in children with chronic liver diseases. J Hepatol. 2010; 53:624–629
3. Ebel NH, Carlin K, Shaffer ML, Shivaram G, Hawkins M, Lane ER, et al. Hepatic venous pressure gradient measurements in children: correlation with hepatic histology and clinical indicators of portal hypertension. J Pediatr Gastroenterol Nutr. 2019; 68:788–792
4. Bass LM, Shneider BL, Henn L, Goodrich NP, Magee JC; Childhood Liver Disease Research Network (ChiLDReN). Clinically evident portal hypertension: an operational research definition for future investigations in the pediatric population. J Pediatr Gastroenterol Nutr. 2019; 68:763–767
5. Shneider BL, de Ville de Goyet J, Leung DH, Srivastava A, Ling SC, Duché M, et al. Primary prophylaxis of variceal bleeding in children and the role of mesorex bypass: summary of the baveno VI pediatric satellite symposium. Hepatology. 2016; 63:1368–1380
6. Malbrunot-Wagner AC, Bridoux L, Nousbaum JB, Riou C, Dirou A, Ginies JL, et al. Transient elastography and portal hypertension in pediatric patients with cystic fibrosis transient elastography and cystic fibrosis. J Cyst Fibros. 2011; 10:338–342
7. Colecchia A, Di Biase AR, Scaioli E, Predieri B, Iughetti L, Reggiani ML, et al. Non-invasive methods can predict oesophageal varices in patients with biliary atresia after a kasai procedure. Dig Liver Dis. 2011; 43:659–663
8. de Franchis R; Baveno VI Faculty. Expanding consensus in portal hypertension: report of the baveno VI consensus workshop: stratifying risk and individualizing care for portal hypertension. J Hepatol. 2015; 63:743–752
9. Fagundes EDT, Ferreira AR, Roquete MLV, Penna FJ, Goulart EMA, Filho PPF, et al. Clinical and laboratory predictors of esophageal varices in children and adolescents with portal hypertension syndrome. J PediatrGastroenterolNutr. 2008; 46:178–183
10. Megremis SD, Vlachonikolis IG, Tsilimigaki AM. Spleen length in childhood with US: normal values based on age, sex, and somatometric parameters. Radiology. 2004; 231:129–134
11. Gana JC, Turner D, Mieli-Vergani G, Davenport M, Miloh T, Avitzur Y, et al. A clinical prediction rule and platelet count predict esophageal varices in children. Gastroenterology. 2011; 141:2009–2016
12. Witters P, Hughes D, Karthikeyan P, Ramakrishna S, Davenport M, Dhawan A, et al. King’s variceal prediction score: a novel noninvasive marker of portal hypertension in pediatric chronic liver disease
. J PediatrGastroenterolNutr. 2017; 64:518–523
13. Voutilainen S, Kivisaari R, Lohi J, Jalanko H, Pakarinen MP. A prospective comparison of noninvasive methods in the assessment of liver fibrosis and esophageal varices in pediatric chronic liver diseases. J Clin Gastroenterol. 2016; 50:658–663
14. Lee MJ, Kim MJ, Han KH, Yoon CS. Age-related changes in liver, kidney, and spleen stiffness in healthy children measured with acoustic radiation force impulse imaging. Eur J Radiol. 2013; 82:e290–e294
15. Cañas T, Maciá A, Muñoz-Codoceo RA, Fontanilla T, González-Rios P, Miralles M, et al. Hepatic and splenic acoustic radiation force impulse shear wave velocity elastography in children with liver disease associated with cystic fibrosis. BioMed Res Int. 2015; 2015:1–7
16. Ferraioli G, Filice C, Castera L, Choi BI, Sporea I, Wilson SR, et al. WFUMB guidelines and recommendations for clinical use of ultrasound elastography: part 3: liver. Ultrasound Med Biol. 2015; 41:1161–1179
17. Pawluś A, Inglot MS, Szymańska K, Kaczorowski K, Markiewicz BD, Kaczorowska A, et al. Shear wave elastography of the spleen: evaluation of spleen stiffness in healthy volunteers. Abdom Radiol (NY). 2016; 41:2169–2174
18. Goldschmidt I, Brauch C, Poynard T, Baumann U. Spleen stiffness measurement
by transient elastography to diagnose portal hypertension in children. J Pediatr Gastroenterol Nutr. 2014; 59:197–203
19. Sutton H, Fitzpatrick E, Davenport M, Burford C, Alexander E, Dhawan A, Grammatikopoulos T. Transient elastography measurements of spleen stiffness as a predictor of clinically significant varices in children. J Pediatr Gastroenterol Nutr. 2018; 67:446–451
20. Burak Özkan M, Bilgici MC, Eren E, Caltepe G. Diagnostic accuracy of point shear wave elastography in the detection of portal hypertension in pediatric patients. Diagn Interv Imaging. 2018; 99:151–156
21. Singh S, Eaton JE, Murad MH, Tanaka H, Iijima H, Talwalkar JA. Accuracy of spleen stiffness measurement
in detection of esophageal varices in patients with chronic liver disease
: systematic review and meta-analysis. Clin Gastroenterol Hepatol. 2014; 12:935–945.e4
22. Ma X, Wang L, Wu H, Feng Y, Han X, Bu H, Zhu Q. Spleen stiffness is superior to liver stiffness for predicting esophageal varices in chronic liver disease
: a meta-analysis. PLoS One. 2016; 11:e0165786
23. Sharma P, Mishra SR, Kumar M, Sharma BC, Sarin SK. Liver and spleen stiffness in patients with extrahepatic portal vein obstruction
. Radiology. 2012; 263:893–899
24. Madhusudhan KS, Kilambi R, Shalimar, Sahni P, Sharma R, Srivastava DN, Gupta AK. Measurement of splenic stiffness by 2D-shear wave elastography in patients with extrahepatic portal vein obstruction
. Br J Radiol. 2018; 91:20180401
25. Gana JC, Turner D, Roberts EA, Ling SC. Derivation of a clinical prediction rule for the noninvasive diagnosis of varices in children. J PediatrGastroenterolNutr. 2010; 50:188–193
26. Adami MR, Ferreira CT, Kieling CO, Hirakata V, Vieira SM. Noninvasive methods for prediction of esophageal varices in pediatric patients with portal hypertension. World J Gastroenterol. 2013; 19:2053–2059
27. Chongsrisawat V, Vejapipat P, Siripon N, Poovorawan Y. Transient elastography for predicting esophageal/gastric varices in children with biliary atresia. BMC Gastroenterol. 2011; 11:41