The aim of this study was to assess the usefulness of picture archiving and communication system (PACS)-based quantitative grayscale ultrasonography (US) measurements in detecting allograft dysfunction in posttransplant patients. In this retrospective study, 116 patients with liver transplantation who underwent biopsy for allograft evaluation were recruited from the database. All participants had US images prior to procedure. Normal, acute cellular rejection (ACR), recurrent hepatitis (Hep), or combined (ACR/Hep) groups were generated based on pathology results. Region of interests were drawn for liver and rectus abdominus muscle to perform quantitative US analysis. The liver/muscle mean ratio (L/M) and heterogeneity index (HI; liver standard deviation/liver mean) were obtained. The ratios of groups were compared, and receiver-operating-characteristic analysis was performed. There was a significant difference between normal (n = 16) and each of other groups (ACR, 39; Hep, 36; combined, 25) for L/M and HI (P < 0.05). No significant difference was detected between ACR, Hep, and combined groups. The areas under the curve for L/M and HI were 0.755 (moderate) and 0.817 (good), respectively. To differentiate abnormal (ACR, Hep, and combined) from normal allografts sensitivity, specificity, PPV, and NPV were 50.0%, 87.5%, 96.2%, and 21.9% for cut point of L/M ≥1 and 84.0%, 68.8%, 94.4%, and 40.7% for cut point of HI ≥0.2 with odds ratios of 7.52 (for L/M ≥1) and 13.10 (for HI ≥0.2), respectively (P < 0.01). L/M has moderate and HI has good discrimination of normal from abnormal allograft in liver transplant patients. PACS-based quantitative US measurements is an objective, easy to use, noninvasive auxiliary tool to discriminate hepatic allograft dysfunction.
Read full abstract