The Liver Segmental Volume Ratio for Noninvasive Detection of Cirrhosis: Comparison With Established Linear and Volumetric Measures.

Abstract

The aim of this study was to compare the liver segmental volume ratio (LSVR), a novel volumetric computed tomography measurement, with established linear measurements for differentiating normal from cirrhotic livers. Hepatic volumes were measured using semiautomated software (Liver Analysis Application, Philips IntelliSpace Portal) on contrast-enhanced abdominal computed tomography scans in 312 adults, including 108 patients with end-stage liver disease (mean age, 55 years; 63 men/45 women) and 204 healthy controls (potential renal donors; mean age, 46 years; 82 men/122 women). The LSVR was defined as the volume ratio of Couinaud segments I to III to segments IV to VIII. Linear measures included the caudate-to-right lobe ratio and maximal splenic dimension. Differences in LSVR between cirrhotics and controls were highly significant (P < 0.0001; mean, 0.55 ± 0.29 versus 0.27 ± 0.07; receiver operating characteristic [ROC] area under the curve [AUC], 0.916). Linear caudate-to-right lobe ratio differences were not statistically significant between the 2 cohorts (P = 0.051; ROC AUC, 0.567). Total liver volume was ineffective for discrimination (ROC AUC, 0.598). An LSVR threshold of 0.35 or greater had a sensitivity and specificity for cirrhosis of 81.5% and 88.7%, respectively. Regional hepatic volume changes, as reflected by the LSVR, are more effective than standard linear measures or total liver volume for differentiating cirrhotic from normal livers.