HepaT1ca - Quantitative Magnetic Resonance Imaging Predicts Individual Future Liver Performance after Liver Resection

Abstract

Introduction: The future liver performance (FLP) of an individual undergoing liver resection for cancer is critical for their survival and recovery. Here, we report the development and clinical testing of HepT1ca, a novel magnetic resonance image (MRI) technology that combines multiparametric MRI signal processing with automated anatomical liver segmentation to estimate FLP. Method: HepaT1ca combines iron-corrected T1 (cT1) mapping with a 3D U-net pipeline to automatically delineate the liver volume and Couinaud segments based on anatomical landmarks. HepaT1ca combines quantitative cT1 mapping with accurate estimation of the future liver remnant (FLR) volume to predict FLP. We evaluated the ability of HepaT1ca to predict post-operative morbidity, length of stay and regenerative capacity in a prospective 2-centre observational clinical trial (ClinicalTrials.gov NCT03213314). Results: 135 of 143 patients recruited and scanned underwent liver resection. 84% of participants had colorectal liver metastases; the remainder had primary liver cancer or other secondary cancers. The HepaT1ca score showed a significant linear correlation with the modified Hyder-Pawlik score, an indicator of post-operative morbidity (adjusted R2=0.26, P< 0.001), and liver regenerative performance (adjusted R2=0.46, P< 0.001). Furthermore, in patients with an FLR < 90%, a mean cT1 >795ms was associated with a longer duration of hospital stay (median (IQR) of 6.5 (5.3-12) vs. 5 (4-7.1); P=0.005). cT1 also correlated with histological measures of inflammation and hepatocyte ballooning. Conclusion: HepaT1ca is a non-invasive quantitative MRI technology for predicting FLP. HepT1ca informs individualised operative risk and augments patient and surgeon decision-making prior to liver resection.