Secondary Logo

Journal Logo

Original Articles: Hepatology

Transient Elastography Measurements of Spleen Stiffness as a Predictor of Clinically Significant Varices in Children

Sutton, Harry∗,†; Fitzpatrick, Emer†,‡; Davenport, Mark; Burford, Charlotte; Alexander, Emma; Dhawan, Anil†,‡; Grammatikopoulos, Tassos†,‡

Author Information
Journal of Pediatric Gastroenterology and Nutrition: October 2018 - Volume 67 - Issue 4 - p 446-451
doi: 10.1097/MPG.0000000000002069


What Is Known

  • Children with clinically significant varices are at increased risk of suffering gastrointestinal bleeding.
  • Transient elastography measurements of liver and spleen stiffness can predict the presence of varices and clinically significant varices in adults.

What Is New

  • Spleen stiffness measurements and liver stiffness measurements can predict clinically significant varices in children with intra- and extrahepatic causes of portal hypertension.
  • Spleen stiffness measurements are stronger predictors of clinically significant varices than previously validated paediatric scores.

Portal hypertension (PHT) results from increased resistance to portomesenteric blood flow and increased intravascular pressure in the portal venous system. In children, most cases occur at the hepatic level with the fibrosis or cirrhosis associated with chronic liver disease (CLD), or at a prehepatic level due to portal vein thrombosis (PVT). Sustained PHT leads to the formation of varices at points of portosystemic venous contact and typically oesophagogastric and rectal varices. In adults, variceal bleeding can be associated with mortality rates of between 10% and 20% in adults and <3% in children (1,2). High-grade varices, also called clinically significant varices (CSV), have a far greater likelihood of bleeding, with 1 study reporting CSV in 96% of children who bled spontaneously (3).

A definitive management protocol for varices in adults has been established from large randomized controlled trials and is defined in the Baveno VI conference (4). In children, the lack of large-scale prospective studies has limited the ability to reach a consensus on variceal management, and variations between centres still exist. Our centre, and several others, has adopted primary prophylaxis during surveillance endoscopies to pre-emptively treat CSV in patients with high-risk of gastrointestinal (GI) bleeding.

A number of noninvasive prediction scores for CSV have been proposed in both adults and children, using a variety of biomarkers such a platelet (PLT) count, spleen size, serum albumin, and liver enzymes. In children, the clinical prediction rule (CPR) using PLTs, spleen size z score, and serum albumin and the King's Variceal Prediction Score (KVaPS), which uses serum albumin and adult equivalent spleen size have been developed (5,6).

Recently, transient elastography (TE), which is a noninvasive tool to estimate liver fibrosis, has been suggested as an ideal method of predicting degree of PHT. In adults, both liver stiffness measurements (LSMs) and spleen stiffness measurements (SSMs) have been able to predict the presence of varices, with the latter being applicable in both CLD and PVT (7–9). TE for LSM and SSM has been shown to be feasible in children and control values in healthy children have been reported (10,11). Paediatric studies have previously shown that LSM is effective in prediction of varices in children with biliary atresia (BA) (12), and SSM is significantly raised in a small number of CLD children with varices, and in those with variceal haemorrhage (11).

The present study aims to determine the feasibility and prognostic value of LSM and SSM by TE in the prediction of CSV in children with PHT of various aetiologies, and compare its prognostic value with other known paediatric prediction scores.



Children with either CLD or PVT reviewed at our paediatric liver centre and selected for endoscopy based on departmental protocol between September 2015 and June 2016 were included. Endoscopy selection criteria were defined as clinically and/or radiologically confirmed splenomegaly and PLT persistently below 100 × 109/L. Children who had received previous endoscopic treatment for gastroesophageal varices were excluded. No children included in the study were treated pharmacologically with beta-blockers before or after endoscopy. Children with GI bleeding were on somatostatin analogues at the time of endoscopy and SSM measurement.

Patient's demographics, laboratory, radiological, and clinical data were recorded at time of endoscopy including sex, age, PLT count, haemoglobin, international normalized ratio, serum albumin, bilirubin, alanine aminotransferase, aspartate aminotransferase, and spleen size measured by ultrasound. Prediction scores, CPR, and KVaPS were also calculated. Data were reported as median and range.

Endoscopic Evaluation

Presence of varices was demonstrated by upper and/or lower gastrointestinal endoscopy. Varices were graded as per UK national guidelines as follows: I- varices that disappear on insufflation of the oesophagus, II- between grade I and III, and III- large varices which occlude the lumen. Additional findings such as gastric varices, graded as either GOV1 or GOV2 based on size, red spots, and portal hypertensive gastropathy were recorded (13,14). CSV were defined as oesophageal grade ≥II and/or gastric varices with stigmata. As per our department protocol all CSV are treated with prophylactic endoscopic therapy in the form of endoscopic sclerotherapy (EST) or endoscopic band ligation (EBL).

Transient Elastography

A trained clinical research assistant performed all TE measurements with a Fibroscanner 502 (Echosens, Paris, France). Both LSM and SSM were attempted on all children included. All TE measurements were recorded within 3 months of endoscopy. The standard 7 mm “M” probe was used on children 6 years or older, whereas the 5 mm child “S1” probe was used for those younger than 6 years, as per manufacturer's recommendation (15). Measurements were recorded by placing the transducer probe in the 10th intercostal space in the right midaxillary line for the liver, and the 10th intercostal space in the left anterior axillary line for the spleen. A valid examination was defined as 10 measurements with a total success rate of ≥90% and the ratio of interquartile range to median value was <30%. The median value of the 10 measurements was reported as the LSM/SSM for the patient, recorded in kilopascals (kPa). TE measurements were not possible in children with excessive abdominal subcutaneous fat or gross ascites, and therefore were not included in the present study.


All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.


Nonparametric univariate analysis was performed using independent t tests for continuous variables and the Chi-squared test for categorical variables. Receiver operating characteristics curve analysis was used to calculate diagnostic accuracy using area under the receiver operating characteristic (AUROC) curve with 95% confidence intervals. The sensitivity, specificity, negative (NPV), and positive predictive values (PPV) were calculated from the best cut-off values defined by the receiver operating characteristics curve analysis, selected to optimize specificity. All statistical analyses were performed using SPSS (IBM, USA, version 23). All comparisons were made using 2-sided significance levels of P < 0.05.


All Children

In total, 67 (32 boys) children, median age 8.5 years (range, 3 months–18 years), underwent TE measurements. Fifty-two (25 boys) had PHT secondary to CLD, whereas 15 (7 boys) had PHT due to PVT. Baseline laboratory data are summarized in Table 1. CSV+ve were observed in 38 (57%). EST was applied in 8 children; the rest (n = 30) had EBL treatment.

Baseline biochemical data, clinical findings, FibroScan measurements, prediction scores, and breakdown of inclusion criteria for all children, chronic liver disease group, and portal vein thrombosis group

There was a significant difference between CSV+ve and CSV-ve subgroups in SSM (53.8 vs 23.9 kPa), but not LSM, CPR, or KVaPS (Table 2). SSM was the only successful predictor of CSV+ve with an optimal cut-off value of 38.0 kPa yielding a sensitivity of 82%, specificity of 89%, PPV of 90%, and a NPV of 80% and an AUROC of 0.86 (Table 3 and Fig. 1).

Mean FibroScan measurements and prediction scores in those with and without clinically significant varices in all children, chronic liver disease, and portal vein thrombosis group
Optimal cut-off value with sensitivity, specificity, positive predictive value, negative predictive value, area under the receiver operator curve, and P value for FibroScan measurements and prediction scores in all children, chronic liver disease group, portal vein thrombosis group, and gastrointestinal bleeding group
Receiver operating characteristics (ROC) of prediction markers for clinically significant varices, including spleen stiffness measurements (SSMs), liver stiffness measurements (LSMs), King's Variceal Prediction Score (KVaPS), and clinical prediction rule (CPR) in all children (i) and children with chronic liver disease (ii).

We added our best performing predictor, SSM, with the other 3 predictors (LSM, CPR, KVaPS) in an attempt to improve its predictive ability for CSV+ve. All 3 were significantly different in all children with CSV+ve compared to CSV-ve (Table 2). Of the combined predictors, SSM combined with LSM was the best at predicting CSV+ve with an AUROC of 0.85, but did not improve upon SSM alone (AUROC = 0.86) (Table 3).

Chronic Liver Disease Group

In the CLD subgroup 52 children (25 boys) median age 8 years (range 3 months–18 years), underwent TE measurements. LSM was successful in 50 (96%) children, whereas SSM was successful in 48 (92%). An unsuccessful TE examination meant we were unable to obtain 10 successful readings satisfying the criteria defined in the methods, reasons for this included large body habitus, children unable to tolerate examination or uncooperative, and BA splenic malformations.

Underlying CLD included BA (n = 33), intestinal failure–associated liver disease (n = 4), cystic fibrosis (n = 3), autoimmune liver disease (n = 3), Wilson disease (n = 3), postliver transplant (n = 2), primary sclerosing cholangitis (n = 2), and other diseases including alpha-1-antitrypsin deficiency and neonatal sclerosing cholangitis (1 each). Six children underwent endoscopy due to GI bleeding (4 episodes of haematemesis, 2 episodes of melaena), the rest as per hospital protocol. CSV+ve were observed in 27 (52%), including all 6 children who suffered a GI bleed. EST was applied in 7 children the rest (n = 20) had EBL treatment.

There was a significant difference between CSV+ve and CSV-ve children in SSM (48.1 vs 16.6 kPa), but not LSM, CPR, or KVaPS (Table 2). SSM proved the best predictor with an optimal cut-off value of 38.05 kPa yielding a sensitivity of 77%, specificity of 87%, PPV of 86%, NPV of 81%, and an AUROC of 0.82 (Table 3 and Fig. 1). There was no improvement in SSM's predictive ability when combined with other scores.

Portal Vein Thrombosis Group

In the PVT subgroup 15 children (7 boys), median age 10 years (range, 2–15 years), underwent TE measurements. LSM and SSM were successful in all 15 children. CSV+ve were observed in 11 (73%) and EST (n = 1) or EBL (n = 10) was used to complete variceal obliteration. SSM was significantly different between the CSV+ve and CSV-ve children (62.8 vs 13.2 kPa), whereas LSM, CPR, and KVaPs were not (Table 2). An SSM cut-off of 16.8 kPa had a predictive ability with 100% sensitivity, 100% specificity, 100% PPV, 100% NPV, and an AUROC of 1.00 (Table 3). There was no improvement in SSM's predictive ability when combined with other scores.

Gastrointestinal Bleeding Group

All children with GI bleeding (n = 6) were from the CLD subgroup. In children with GI bleeding there was a significant difference in LSM (46.8 vs 22.0 kPa), KVaPS (56.1 vs 76.8), and CPR (90.9 vs 106.9), but not SSM when compared with those who did not suffer a GI bleed. For the prediction of GI bleeding LSM performed the best with an optimal cut-off value of 34.3 kPa yielding a sensitivity of 80%, specificity of 88%, PPV of 22%, NPV of 96%, and an AUROC of 0.84 (Table 3). Two children in the study who had not bled previously developed GI bleeding on follow-up, both more than 6 months from TE measurements. Each child had undergone endoscopy when originally scanned, both were in the CSV+ve group and had received prophylactic EST. Both children had SSM and LSM measurements above the cut-offs for CSV+ve and GI bleeding (45.7, 42.8 and 71.0, 35.3 kPa, respectively).


TE studies have been previously demonstrated to be feasible and reproducible in children (10,15,16). Our study had success rates of 97% and 94% for LSM and SSM, respectively. Our results showed that out of the studied predictors SSM, LSM, KVaPS, and CPR, only SSM was successfully able to stratify children based on the presence of CSV. Of the 4, SSM, with an AUROC of 0.92, was best predictor in all children, and exclusively in intra- and extrahepatic causes of PHT. Combining SSM with the other predictors did not improve its predictive ability. At a cut off of 38.0 kPa SSM had an 80% NPV, and by applying this cut-off in clinical practice we could have prevented 80% of endoscopies in which no treatment was offered.

It is interesting that neither CPR nor KVaPS were able to successfully predict the presence of CSV+ve. Although CPR was designed using children with both intra- and extrahepatic causes of PHT, it was developed for the prediction of any varices and not specifically CSV+ve, which may explain its failure to predict CSV+ve in our study (5). Although KVaPS was developed to predict CSV+ve, it was designed only for children with CLD. Also, the design study included children who did not undergo endoscopy, as they did not meet department criteria, and were assumed to be CSV-ve, as oppose to the present study in which only children with endoscopically confirmed findings were included (6).

It should be noted that the FibroScan model used in the present study had a maximum recording of 75.0 kPa. In our cohort 7 children (4 PVT and 3 CLD, all CSV+ve) had recordings of 75.0 kPa (Supplemental Fig. 1, Supplemental Digital Content, Although we cannot determine their exact TE measurements, as they are all ≥75.0 kPa it would not have an effect on the reported medians or cut-off values.

As mentioned above, our centre uses a combination of splenomegaly and decreased PLT to select children for endoscopy. This method has been previously validated and is widely used in paediatric hepatology centres that perform surveillance endoscopies (17). In our study, this screening test identified 68 children, during our enrolment period, to undergo endoscopy to prophylactically treat CSV. Of the 68 children selected only 38 (57%) had CSV+ve in whom prophylactic treatment was offered. A pre-endoscopy screen using the above mentioned SSM cut-off would have prevented 25 endoscopies in which no treatment was offered and would have minimized the number of endoscopies performed. Our centre is currently offering TE SSM and LSM routinely, in our CLD with hypersplenism and PVT population. We continue offering simultaneous OGD and SSM to our patients but we also monitor SSM measurements during their clinic follow-up in between endoscopies for PHT progression purposes.

Of note, SSM was the only marker that was not significantly different in those with GI bleeding. This is in conflict with a previous study, which showed that SSM measurements are significantly raised in children with GI bleeding (9). It is noteworthy that despite being the best predictor of CSV SSM was unable to successfully predict GI bleeding but underpins the complex pathophysiology of GI bleeding involving potential dysregulation in metabolic, endothelial, or coagulation pathway (18). We believe evaluating SSM's usefulness in predicting GI bleeding is beyond the aims of the study as the results from the GI bleeding group are limited by the small sample size (n = 6) and consisting entirely of the CLD subgroup. Furthermore, the 2 children who bled during follow-up both had CSV and high SSM measurements (45.7 and 42.8 kPa) at the time of their initial endoscopy, and underwent prophylactic EST. The findings of the present study suggest the need for more research into the relationship between spleen stiffness and gastroesophageal varices. A study in adults with cirrhosis demonstrated that during 5-year follow-up spleen enlargement was associated with variceal formation and growth (19). A longitudinal study is needed to explore the relationship between spleen stiffness and variceal growth over time, and potential acute changes to spleen stiffness around the time of a variceal bleed.

Surveillance endoscopies and primary prophylaxis on children is considered controversial and has not been endorsed by the Baveno VI conference due to insufficient evidence supporting its safety and efficacy (20). Other prophylactic measures to prevent GI bleeding used in adults, such as nonselective beta-blockers, are also not fully endorsed in paediatrics (7). Because of the increased risk of GI bleeding that CSV pose, some paediatric liver units, including our centre, monitor children with PHT with regular surveillance endoscopies and prophylactically treat CSV with endoscopic therapy. A recently published study (3) reviewed 1300 children during a 25-year period and showed that primary endoscopic prophylaxis was a safe and effective means of managing PHT in children. That same study found that in children who did not receive primary prophylaxis 96% of those who bled spontaneously had CSV at the time of the bleed.

The need for a clinically applicable, noninvasive marker of PHT has led to a number of studies, like ours, looking to find the best noninvasive predictor of PHT (21). Although the majority of this research has been done in adults, the 2 prediction scores tested in this study, CPR and KVaPS, were developed in paediatric studies (5,6).

Our study looked to expand on the work done developing PHT prediction scores for children. The CPR was developed to predict the presence of any varices in children, whereas KVaPS was developed for children with PHT secondary to CLD only (8,9). A recent study examined shear wave elastography's ability to predict clinically significant PHT in children as determined by hepatic venous pressure gradient measurements (22). Our study differs in that it is, to our knowledge, the first large-scale observational study assessing the ability of TE measurements to predict the presence of CSV by measuring LSM and SSM, in children with intra- and extrahepatic causes of PHT, and compare it with previously validated prediction scores.

Despite SSM's success at predicting CSV, TE measurements have limitations. TE is difficult to perform in patients with obesity, and is contraindicated in those with ascites, a common complication of CLD. This limitation may be overcome by assessing spleen stiffness with magnetic resonance technology. A study found that magnetic resonance elastography (MRE) of spleen and liver stiffness were both able to successfully predict CSV in adults with cirrhosis (23). MRE in children may be able to provide similar predictive ability as TE, possibly with fewer restrictions in its applicability but MRE is a nonbedside test with children often requiring sedation or a general anaesthetic.

Managing PHT in children with surveillance is invasive, expensive, and a disruption to family life, necessitating inpatient stay. The long-term aim of the present study is to minimize the number of unnecessary endoscopies we perform by determining the best way to stratify children based on the presence of CSV. The findings of our study demonstrate that TE measurements of liver and spleen stiffness are a noninvasive test, which could be used at regular follow-up outpatient appointments to stratify children who would benefit from endoscopy, allowing the prevention of GI bleeds, while reducing the number of unnecessary endoscopies.


1. Kim SJ, Kim KM. Recent trends in the endoscopic management of variceal bleeding in children. Pediatr Gastroenterol Hepatol Nutr 2013; 16:1–9.
2. Jafar W, Jafar AJN, Sharma A. Upper gastrointestinal haemorrhage: an update. Frontline Gastroenterol 2016; 7:32–40.
3. Duche M, Ducot B, Ackermann O, et al. Portal hypertension in children: high-risk varices, primary prophylaxis and consequences of bleeding. J Hepatol 2017; 66:320–327.
4. De Franchis R, Baveno VIF. 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.
5. Gana JC, Turner D, Roberts EA, et al. Derivation of a clinical prediction rule for the noninvasive diagnosis of varices in children. J Pediatr Gastroenterol Nutr 2010; 50:188–193.
6. Witters P, Hughes D, Karthikeyan P, et al. King's Variceal Prediction Score: a novel noninvasive marker of portal hypertension in pediatric chronic liver disease. J Pediatr Gastroenterol Nutr 2017; 64:518–523.
7. Berzigotti A. Non invasive evaluation of portal hypertension using ultrasound elastography. J Hepatol 2017; 67:399–411.
8. Singh S, Eaton JE, Murad MH, et al. 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.e4–945.e4.
9. Sharma P, Mishra SR, Kumar M, et al. Liver and spleen stiffness in patients with extrahepatic portal vein obstruction. Radiology 2012; 263:893–899.
10. Engelmann G, Gebhardt C, Wenning D, et al. Feasibility study and control values of transient elastography in healthy children. Eur J Pediatr 2012; 171:353–360.
11. Goldschmidt I, Brauch C, Poynard T, et al. Spleen stiffness measurement by transient elastography to diagnose portal hypertension in children. J Pediatr Gastroenterol Nutr 2014; 59:197–203.
12. Chongsrisawat V, Vejapipat P, Siripon N, et al. Transient elastography for predicting esophageal/gastric varices in children with biliary atresia. BMC Gastroenterol 2011; 11:41.
13. Sarin SK, Lahoti D, Saxena SP, et al. Prevalence, classification and natural history of gastric varices: a long-term follow-up study in 568 portal hypertension patients. Hepatology 1992; 16:1343–1349.
14. Jalan R, Hayes PC. UK guidelines on the management of variceal haemorrhage in cirrhotic patients. British Society of Gastroenterology. Gut 2000; 46 (suppl 3–4):III1–III15.
15. Goldschmidt I, Streckenbach C, Dingemann C, et al. Application and limitations of transient liver elastography in children. J Pediatr Gastroenterol Nutr 2013; 57:109–113.
16. Fitzpatrick E, Quaglia A, Vimalesvaran S, et al. Transient elastography is a useful noninvasive tool for the evaluation of fibrosis in paediatric chronic liver disease. J Pediatr Gastroenterol Nutr 2013; 56:72–76.
17. Gana JC, Turner D, Mieli-Vergani G, et al. A clinical prediction rule and platelet count predict esophageal varices in children. Gastroenterology 2011; 141:2009–2016.
18. Sutton H, Dhawan A, Grammatikopoulos T. Non-invasive markers of portal hypertension: appraisal of adult experience and potential utilisation in children. J Pediatr Gastroenterol Nutr 2018; 66:559–569.
19. Berzigotti A, Zappoli P, Magalotti D, et al. Spleen enlargement on follow-up evaluation: a noninvasive predictor of complications of portal hypertension in cirrhosis. Clin Gastroenterol Hepatol 2008; 6:1129–1134.
20. Shneider BL, De Ville de Goyet J, Leung DH, 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.
21. Deng H, Qi X, Guo X. Diagnostic accuracy of APRI, AAR, FIB-4, FI, King, Lok, Forns, and FibroIndex scores in predicting the presence of esophageal varices in liver cirrhosis: a systematic review and meta-analysis. Medicine (Baltimore) 2015; 94:e1795.
22. Yoon HM, Kim SY, Kim KM, et al. Liver stiffness measured by shear-wave elastography for evaluating intrahepatic portal hypertension in children. J Pediatr Gastroenterol Nutr 2017; 64:892–897.
23. Shin SU, Lee JM, Yu MH, et al. Prediction of esophageal varices in patients with cirrhosis: usefulness of three-dimensional MR elastography with echo-planar imaging technique. Radiology 2014; 272:143–153.

clinically significant varices; portal hypertension; spleen stiffness; transient elastography

Supplemental Digital Content

Copyright © 2018 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition