Evaluating agreement between solid tumor measurements used to assess response

Abstract

Objective: To propose and demonstrate using Bland-Altman plots and Limits of Agreement based on the relative difference(RD) in solid tumor measurements to assess agreement.

Methods: A modification to the Bland-Altman plot which involves replacing the difference plotted on the vertical axis withthe relative percent difference between two measurements of tumor burden is discussed. Quantifying tumor burden requiressumming skewed individual tumor measurements. This quantity is the same one used in assessing tumor response to therapeuticagents and is familiar to radiologists and clinicians working with cancer patients. The distribution of the relative difference isexplored and revised equations for the limits of agreement are presented. The methods are then applied to positron emissiontomography data studying two radiotracers.

Results: The distribution of the relative difference is highly skewed and can be approximated by a negative shifted lognormaldistribution. The limits of agreement for the RD need to appropriately reflect this distribution. The standard equations for the95% limits of agreement assume the differences are approximately normally distributed and may not be appropriate for therelative difference.

Conclusions: The modified Bland-Altman plot is based on a clinically meaningful quantity and provides a method for handlingdata with multiple lesions per patient.