Performance of spleen stiffness measurement by 100‐Hz vibration‐controlled transient elastography, liver stiffness, APRI score and their combination for predicting oesophageal varices in liver cirrhosis

Abstract

BackgroundOesophageal varices (EV) rupture remains one of the most severe complications of cirrhosis. As the gold standard to predict this accident, esophagogastroduodenoscopy (EGD) itself also has a weakness. Not all patients are convenient with this modality in clinical practice apart from the risk and cost burden. Hence, the search for other non-invasive modalities with high accuracy is still noteworthy. Among them, spleen stiffness measurement (SSM) with 100 Hz probe, liver stiffness measurement (LSM), and the aspartate amino transferase to platelet ratio index (APRI) score became popular and intensively studied with good accuracy, but the results remain conflicting. This study aims to investigate the performance of SSM, LSM, APRI score, and their combination especially as a screening tool for predicting EV in liver cirrhosis patients. MethodsIn this cross-sectional study, we included 141 patients with liver cirrhosis who had undergone endoscopy, SSM, LSM, and APRI score calculation between January and March 2023 were enrolled. Diagnostic accuracy was assessed by the area under the receiver-operator curve (AUC). Transient elastography (TE) measurement was performed using a spleen-dedicated FibroScan with a 100-Hz probe. ResultsOf the 141 patients, the most common aetiology was hepatitis B in 71 patients (50.4 %). EV were found in 116 patients. Using the AUC, SSM at a cutoff of 40 kPa had the best performance with an AUC of 0.892 (CI 95 %: 0.814–0.969, p <0.0001), with sensitivity 88.79 % and specificity 80 %). Meanwhile, LSM and APRI score had an AUC of 0.832 (CI 95 %: 0.742–0.922, p <0.0001) and 0.780 (CI 95 %: 0.660–0.900, p <0.0001), respectively. The combination of all measurement tools did not show better performance than SSM alone with an AUC of 0.892 (CI 95 %: 0.802–0.982, P <0.0001) ConclusionSSM provides better performance than LSM and APRI scores for predicting EV. Performance of SSM alone is non-inferior compare to multiple diagnostic tools combined.