Meghan G. Lubner MD, Presenter: Nothing to Disclose

Nicholas Stabo, Abstract Co-Author: Nothing to Disclose

Sam Lubner, Abstract Co-Author: Nothing to Disclose

Alejandro Munoz Del Rio PhD, Abstract Co-Author: Research Consultant, Cellectar Biosciences, Inc Reviewer, Wolters Kluwer nv

Chihwa Song PhD, Abstract Co-Author: Nothing to Disclose

Perry J. Pickhardt MD, Abstract Co-Author: Co-founder, VirtuoCTC, LLC Stockholder, Cellectar Biosciences, Inc

To compare reproducibility of unidimensional and volumetric measures of metastatic colorectal cancer by reader, site of disease and time point in disease assessment

Analysis of CT images in 107 patients (mean age 58.7, range 25-81; 47 F, 60 M) undergoing systemic treatment for metastatic colorectal cancer was performed. Unidimensional (1D) and volumetric (3D) measures were retrospectively obtained on index lesions at three time points (mean interval 4.1 mos, median 3.7 mos) by three readers (abdominal imager, imaging core-lab manager, medical student) using a semi-automated technique. Measurements were summed and compared using best overall response and response at first post-treatment time point. Patient response was categorized based on RECIST 1.1 (CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease). Intra and interobserver variability of 1D and 3D measures was assessed. Reproducibility of measurement by metastatic lesion site was evaluated. Kaplan-Meier models for each with categorical tumor response were constructed and compared. These models were created both for “best overall response” and for response at the first post treatment time point.  

Volumetric measures showed similar intra and interobserver variability to unidimensional measures. Metastatic site (lung, liver, node, other) did not significantly impact measurement reproducibility. Kaplan-Meier curves for unidimensional vs volumetric assessment were very similar in appearance. Both 1D and 3D measurements separated PD from the SD/PR group, but neither separated SD and PR well. Similar KM curves were seen using overall best response (across all time points) compared to response at the first post treatment time point. 

Volumetric assessment of metastatic CRC is fairly reproducible, but does not show improved ability to predict survival over 1D measures and may not represent an improvement over less complex standard linear measures for this indication.  Categorization of these variables either using best overall response or at the first post treatment time point also show similar survival models.

3D measures are reproducible, but do not improve correlation with survival over 1D measures.  Assessment of disease at the first time point correlates well with overall best response in metastatic CRC, which may enable transition of therapy earlier.

Lubner, M, Stabo, N, Lubner, S, Munoz Del Rio, A, Song, C, Pickhardt, P, Volumetric Assessment of Metastatic Colorectal Cancer: Reproducibility by Reader, Site, and Time Point.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14016369.html