Noninvasive Rapid Acquisition Technique for Predicting Esophageal Varices Using Magnetic Resonance Elastography

Abstract

Introduction: Current guidelines recommend serial variceal screening of cirrhotic patients with esophagogastroduodenoscopy (EGD). Evidence does not support any noninvasive method. Magnetic Resonance Elastography (MRE) and labs are noninvasive modalities used to help predict hepatic fibrosis. We evaluated these methods as potential predictors for the detection of esophageal varices (EV). Methods: Twelve patients with clinically compensated cirrhosis undergoing an EGD for variceal screening or surveillance were recruited for the study (six without EV and six with EV). Patients with a history of variceal ligation or currently on a non-selective beta blocker at the time of the EGD were excluded. MRE was then prospectively performed using a 3T MRI scanner (Tim-trio, Siemens Healthcare, Germany) using standard (MREs) and rapid (MREr) sequences. MRE wave images were processed using online reconstruction to report a mean stiffness value (kPa) for the liver as well as the spleen. Routine laboratory data was collected to calculate the model for end-stage liver disease (MELD) scores, AST to platelet ratio index (APRI) and The Lok index. Univariate logistic regression models were used to determine the odds of a patient having EV for a one unit increase in each of the variables. Results: The odds ratio (OR) for having EV for an increase of one unit in the liver stiffness score is 0.633 using MREs and 0.698 using MREr. The OR for having EV for an increase of one unit in the spleen stiffness score is 0.903 using MREs and 0.735 using MREr. Increases in liver and spleen stiffness scores for both MREs and MREr techniques were associated with lower odds of predicting EV, however, statistical significance was not achieved (p values ranged from 0.323-0.753). The OR for having EV for an increase of one unit in MELD is 1.254, APRI is 3.597, and Lok index is 1.080. Increases in all scores were associated with higher odds of predicting EV, however, statistical significance was not achieved (p values ranged from 0.227-0.922). Conclusion: Our experience thus far has shown that neither the use of MRE stiffness values nor other models predictive of hepatic fibrosis are accurate in screening for the presence of EV, but our small sample size has limited any further statistical analysis. MRE is a safe and effective method for the assessment of liver fibrosis, and further investigation for its role in conjunction with other predictive models is warranted to find a potential noninvasive screening method for EV.