For 30 years, the standard way to assess a patient's response to treatment has been to monitor changes in tumor diameter using serial CT exams. However, studies began to show the inadequacy of conventional methods in assessing tumor responses to novel targeted therapies. Indeed, our recent study correlating early radiographic response induced by a targeted therapy, gefitinib, with the presence of epidermal growth factor receptor mutations suggests that tumor volume change is better for detecting biologic activity of the target in non-small cell lung cancer. Despite the widespread use of CT in response assessment, contemporary analysis of CT data and its role in guiding clinical trial interpretation has neither been validated nor optimized. It is now critical that such a rigorous evaluation be undertaken as clinical trials begin incorporating volumetric CT (VCT) into the assessment of new therapies and as more computer algorithms haven been/are being developed for assisting in measuring tumor volumes. Our first goal is to systematically explore sources and variability in volume and volume change measurements that are introduced during image acquisition and tumor size measurement using both an FDA chest phantom with known ground truth (volume) and in vivo lung cancer tumors taken on same-day repeat CT scans (Aims 1 and 2). Achieving this goal will allow us to gain key insights into how and to what extent different CT vendor platforms and scanning parameters, serial CT scans, and image segmentation algorithms affect tumor volume measurement and subsequent response assessment. Such information will be essential for helping establish standards for CT imaging protocols and select appropriate algorithms to improve accuracy, consistency of volume measurements and response assessments in multicenter clinical trials. Even with improvements in accuracy and reproducibility, enhanced image quantification alone will not be accepted by the oncology community as an improved biomarker for novel therapies without showing a robust correlation of VCT with clinical outcomes. Our second goal, therefore, is to validate VCT as an early and more accurate biomarker for predicting pathologic response and disease free survival. We will do this in a prospective multicenter lung cancer trial (CALGB 30803) that is evaluating a novel neoadjuvant regimen (Aim 3). Because CT imaging technology is globally available, our algorithm can be widely distributed, run on standard computers and be integrated with picture archiving and communication systems, we expect that our findings will be broadly useful for both drug discovery and clinical care.
Confirmation of the ability of VCT to detect earlier response or progression would lower the cost and accelerate the timelines of future clinical trials. The outcome of this study will be of great value in helping establish standards of imaging protocols and response criteria for the use of VCT in clinical trials and clinical care where treatment benefit is measured by the changes in tumor size assessed by CT.
|Aerts, Hugo J W L; Grossmann, Patrick; Tan, Yongqiang et al. (2016) Defining a Radiomic Response Phenotype: A Pilot Study using targeted therapy in NSCLC. Sci Rep 6:33860|
|Kalpathy-Cramer, Jayashree; Zhao, Binsheng; Goldgof, Dmitry et al. (2016) A Comparison of Lung Nodule Segmentation Algorithms: Methods and Results from a Multi-institutional Study. J Digit Imaging 29:476-87|
|Zhao, Binsheng; Tan, Yongqiang; Tsai, Wei-Yann et al. (2016) Reproducibility of radiomics for deciphering tumor phenotype with imaging. Sci Rep 6:23428|
|Zhao, Binsheng; Lee, Shing M; Lee, Hyun-Ju et al. (2014) Variability in assessing treatment response: metastatic colorectal cancer as a paradigm. Clin Cancer Res 20:3560-8|
|Zhao, Binsheng; Tan, Yongqiang; Bell, Daniel J et al. (2013) Exploring intra- and inter-reader variability in uni-dimensional, bi-dimensional, and volumetric measurements of solid tumors on CT scans reconstructed at different slice intervals. Eur J Radiol 82:959-68|
|Tan, Yongqiang; Schwartz, Lawrence H; Zhao, Binsheng (2013) Segmentation of lung lesions on CT scans using watershed, active contours, and Markov random field. Med Phys 40:043502|