Measurement of multiple spleen lengths is not necessary for non-invasive prediction of high-risk esophagogastric varices

Aim To validate an appropriate spleen size measurement technique for the prediction of high-risk esophagogastric varices. Methods This retrospective cross-sectional study included 369 patients who underwent ultrasonography and computed tomography (CT) of the spleen and esophagogastroduodenoscopy between January 2018 and December 2020. Maximum spleen length, width, and craniocaudal length were measured in a longitudinal view. The two-dimensional (2D) spleen index (maximum length x maximum width in the longitudinal view) was calculated. A three-dimensional (3D) spleen index was then defined as follows: 2D spleen index x maximum length in the transverse view. The similarity in spleen volume measured by CT and ultrasonography (spleen index) was assessed by the correlation coefficient. The diagnostic accuracies of the spleen index, platelet/spleen length, and platelet/spleen index were calculated to determine the overall diagnostic accuracy. Results Compared to the other spleen indices, our 3D spleen index was significantly better correlated with spleen volume on CT (r = 0.91, 95% confidence interval 0.89-0.92, p < 0.001). Receiver-operating characteristic curve analyses revealed no significant difference between the 3D and 2D indices (p = 0.228) but did show a significant difference between the 3D and one-dimensional indices (p = 0.020). Although the area under the curve for the platelet count combined with the spleen index or length was higher than that for our 3D index, there was no significant difference between platelet count and spleen index or length (p = 0.078). Conclusions Platelet/spleen length has a reasonable ability to predict high-risk esophagogastric varices, even though measurement of two or three factors can be correlated with spleen volume.