Limitations of GD-EOB-DTPA-enhanced MRI: can clinical parameters predict suboptimal hepatobiliary phase?

Abstract

Aim: To establish cut-off levels of the clinical parameters, which would predict suboptimal 30 minutes delayed hepatobiliary phase (HBP) with high specificity.

Materials and methods: This retrospective study included patients with chronic liver disease who underwent hepatocellular carcinoma screening with Gd-EOB-DTPA-enhanced magnetic resonance imaging (MRI) between 1 January 2011 and 30 November 2014. For each case, HBP was graded as adequate or suboptimal, based on Liver Image Reporting and Data System (LI-RADS) criteria. The following laboratory data obtained within 3 months of the MRI date was extracted: total bilirubin (TB), direct bilirubin (DB), serum glutamic oxaloacetic transaminase (SGOT), serum glutamic-pyruvic transaminase (SGPT), alkaline phosphatase (ALP), albumin, activated partial thromboplastin time (aPTT), and International normalised ratio (INR). Model For End-Stage Liver Disease (MELD) scores were calculated as 3.78×ln[TB] + 11.2×ln[INR] + 9.57×ln[creatinine] + 6.43. Receiver operating characteristic (ROC) curve analysis was used to establish cut-off values for predicting suboptimal HBP.

Results: Of 284 patients, 242 (85.2%) patients (91; 57.6% male) had an adequate HBP and 42 (14.8%) patients (13; 61.9% male) had suboptimal HBP, with mean ages of 58.5±9.7 years and 55±12.7 years, respectively (p=0.096). Areas under the ROC curve for predicting suboptimal HBP were 0.85 (95%CI 0.79-0.91) for the MELD score, 0.88 (95%CI 0.82-0.93) for TB, and 0.91 (95%CI 0.86-0.95) for DB. Accuracy, positive likelihood ratios and cut-off values for predicting suboptimal HBP were, respectively: 86.7% and 11.2 for the MELD score ≥16.7, 88.2% and 28.7 for TB ≥4.3 mg/dl, and 91.1% and 36.4 for DB ≥1.3 mg/dl. SGOT, SGPT, and ALP were not statistically significantly different between the groups.

Conclusion: Cut-off levels of MELD score, DB, and TB can predict an suboptimal HBP with high accuracy. Prospective identification of patients with a high likelihood of an suboptimal HBP can help to avoid administering a more costly agent to patients who would not benefit from its unique properties.