Accuracy of splenic stiffness measurement for predicting clinically significant varices in children with portal hypertension: a systematic review and diagnostic test accuracy meta-analysis.

Abstract

Objective: To evaluate the diagnostic accuracy of splenic stiffness measurement (SSM) for detecting clinically significant varices (CSV) in children with portal hypertension (PHTN), including non-cirrhotic portal hypertension (NCPH) and chronic liver disease (CLD).

Design: Systematic review and diagnostic test accuracy meta-analysis.

Data sources: MEDLINE (via PubMed), Embase and Scopus were searched from inception to 31 July 2025. No language or time limitations were applied.

Eligibility criteria: We included prospective and retrospective studies which had children (<18 years) with clinically, radiologically, biochemically or histologically diagnosed PHTN. The index test was SSM (transient elastography, 2D-shear wave elastography (SWE) or point-SWE/acoustic radiation force impulse (ARFI)). The reference standard was oesophagogastroduodenoscopy. Studies with other reference standards, like liver biopsy, were excluded.

Data extraction and synthesis: Two reviewers independently extracted data. A Bayesian random-effects bivariate model was used to estimate summary sensitivity and specificity with 95% credible intervals (CrIs), and hierarchical summary receiver operating characteristic curves were constructed. Risk of bias was assessed using QUADAS-2, and certainty of evidence (COE) was evaluated using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) approach.

Results: 21 studies were included in the systematic review and 14 (n=1027) were eligible for meta-analysis. For CLD (8 studies, n=405), summary sensitivity was 83.0% (95% CrI 69.9% to 91.0%) and specificity was 75.7% (60.8%-85.2%), showing low certainty for CSV. In NCPH (6 studies, n=171), sensitivity for CSV was 92.5% (82.7%-97.0%) and specificity was 80.3% (57.6%-92.7%), with moderate COE for sensitivity and low COE for specificity.

Conclusions: SSM probably has high sensitivity and may have moderate specificity for detecting CSV in children with NCPH. In CLD, diagnostic accuracy remains uncertain due to low COE. Further well-designed, aetiology-specific studies are needed to establish the role of SSM in the clinical pathway.

Prospero registration number: CRD42024579988.