Measuring prostate cancer lesion size on serial MRI during active surveillance: A multi-centre observational study

To assess the intra- and inter-reader reliability of measurements of prostate cancer (PCa) lesions on serial multiparametric MRI scans, and to assess how lesion characteristics and measurement methods influence measurements over time in an international two-centre setting.

Twenty patients with low- or intermediate-risk PCa undergoing active surveillance were included. A total of three MRI scans were evaluated for each patient: one baseline and two serial follow-up scans. Five consultant radiologists from two institutions independently measured the size of the index lesion at each time point by four methods: maximum diameter, biaxial diameters, ellipsoid volume, and planimetry. Measurements were repeated after a four-week 'wash-out' period. Variability was assessed using strip plots and Bland-Altman analysis.

Graphical presentation of inter-reader and intra-reader reliability enabled examination of variability between methods, readers and reads. Variability increased with lesion size and was greatest for lesions with poorly defined margins. Smaller, well-delineated lesions demonstrated greater consistency over time and across readers. Estimates of percentage change showed greater variability than absolute measurements, particularly for volumetric methods. Small differences in the underlying absolute measurements are amplified when expressed relative to baseline values.

Serial MRI assessment of PCa lesion size is subject to substantial variability, particularly for larger lesions and lesions with indistinct boundaries. The benefits of consistent readers and measurement methods warrant further confirmation in larger prospective studies to support reliable clinical decision-making during active surveillance.