CT Colonography in Preoperative Staging of Colon Cancer: Evaluation of FOxTROT Inclusion Criteria for Neoadjuvant Therapy
Natally Horvat1,2,3, Aradhna Raj1, Sandy Liu4, Kristina A. Matkowskyj5, Andrea Knezevic6, Marinela Capanu6, Jinru Shia4, Perry J. Pickhardt7 and Marc J. Gollub1
OBJECTIVE. The purpose of the present study is to evaluate the diagnostic performance of and interreader agreement for CT colonography (CTC) in the local staging of colon cancer, with emphasis given to the FOxTROT (Fluoropyrimidine, Oxaliplatin, and Targeted-Receptor pre-Operative Therapy [Panitumumab]) trial inclusion criteria, which propose a new tailored treatment paradigm for colon cancer that uses neoadjuvant therapy for patients with a high-risk of locoregional disease as determined by imaging.
MATERIALS AND METHODS. This biinstitutional retrospective study involved 89 patients (with 93 tumors) who had colon cancer and underwent presurgical CTC. Two radiologists reviewed the CTC studies for local staging, including measurement of the tumor beyond the muscularis propria on a true orthogonal plane. Histopathologic findings for surgical colectomy specimens served as the reference standard for local pathologic staging. The sensitivity, specificity, positive predictive value, and negative predictive value for local determination of the T category, N category, and extramural vascular invasion (EMVI) were calculated separately for each reader. High-risk T category tumors were the same as those as used in the FOxTROT trial. Interreader agreement was assessed using the kappa statistic.
RESULTS. Thirty-five of 93 tumors (37.6%) were histologically classified as high-risk tumors (T3c, T3d, or T4 tumors). The interreader agreement was substantial (κ = 0.68) for classifying high-risk tumors with the use of CTC, moderate for differentiating N0 from N1 and N2 (κ = 0.44), and slight for detecting EMVI (κ = 0.15). The diagnostic statistics for CTC for the two readers were as follows: for detection of high-risk tumors, sensitivity was 65.7% and 82.9%, and specificity was 81.0% and 87.9%; for detection of N category–positive disease, sensitivity was 50.9% and 69.8%, and specificity was 50.0% and 72.5%; and for detection of EMVI, sensitivity was 18.2% and 66.7%, and specificity was 60.0% and 91.7%.
CONCLUSION. The present study shows that CTC might be a feasible imaging modality for preoperative local staging of higher-risk colon cancers for which neoadjuvant chemotherapy is more suitable on the basis of the FOxTROT trial criteria. However, further studies are required to allow a better generalization of our results.