Improving the accuracy of MRI spleen and liver volume measurements: A phase III Gaucher disease clinical trial setting as a model

Purpose: To achieve minimal inter-observer variability in assessment of spleen and liver volume changes using a novel MRI reading method in the context of a phase III clinical trial of a new therapy for Gaucher disease. Materials and methods: Abdominal MRI examinations at screening and after 6 and 9 months' exposure to a novel plant-cell-derived recombinant enzyme, taliglucerase alfa, were taken in 31 patients with Gaucher disease and at least 8-fold greater than expected splenomegaly. Transverse 12, 11, and in/out-of-phase, and coronal T1 sequences were performed using standardized settings across 11 sites globally. Spleen and liver volumes were semi-automatically delineated using an automatic segmentation algorithm followed by manual correction by experienced technologists using advanced editing tools. Data of all randomized patients were then submitted for efficacy evaluation to two independent experts blinded to time-point and treatment. Results: Mean (+/- SD) percent variability over all time-points was 030%+/- 0.46% for spleen and 0.53%+/- 0.69% for liver using 178 spleen and liver volumes measured twice. Adjudication due to >= 5% variability between observers was not required. Conclusion: The measurement method was found to be precise in monitoring spleen and liver volume changes over time, with a much lower variability than traditional manual methods, supporting the accuracy of the results. Given the observed minimal variability rates among multiple readers, a single read of each volume would be sufficient. (C) 2010 Elsevier Inc. All rights reserved.