Consistency of Radiological Staging in Resectable Mismatch-Repair Proficient Colon Cancer: An Interobserver Agreement
Vincenzo Nasca1, Gabriele Tinè2, Marta Vaiani3
1Department of Medical Oncology, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy.
Background:
Reproducibility and accuracy of radiological classification with CT imaging is crucial for selecting patients with resectable, nonmetastatic colon cancer (CC) who may benefit from neoadjuvant strategies.
Methods:
We evaluated interobserver agreement among 4 independent radiology equipes in classifying patients with resectable, nonmetastatic, mismatch repair proficient CC with available preoperative CT imaging. Radiology equipes were blinded to clinical and pathological data, and categorized tumors according to FOxTROT and Node-RADS criteria, classifying T (T1/2, T3, T4, including stratification by extramural invasion [EMI]) and N status. The primary endpoint was interobserver agreement (Cohen's κw); secondary endpoints included concordance with pathological staging and diagnostic performance metrics. Bayesian and cross-validation analyses assessed robustness.
Results:
Among 109 patients (32.1% pT2, 33.9% pT3, 33.9% pT4; 45.9% pN0/1c, 54.1% pN1-2), overall interobserver agreement indicated inadequate concordance for both cT (κw = 0.56; 95% CI, 0.43-0.68) and cN category (κw = 0.51; 95% CI, 0.36-0.66). Of 109 patients, 38 (35%) were classified identically for T stage by all 4 Radiology groups, while 74 (68%) showed some disagreement. When considering risk stratification into high risk (cT3 with EMI ≥5 mm or cT4) versus low risk (cT1/cT2 or cT3 with EMI <5 mm) 56 patients (51%) received the same classification all 4 groups, and 95 patients (87%) were classified consistently by at least 3 groups. Overall concordance between radiology and pathological staging was inadequate for both T (κw = 0.47; 95% CI, 0.35-0.57) and N staging (κw = 0.16; 95% CI, 0.07-0.33). The mean sensitivity and specificity of CT scan for pT3/T4 and pT4 were 72%/78% and 33%/93%.
Conclusions:
Identification of high-risk features in resectable CC by CT scan should be implemented to guide patients' selection for neoadjuvant therapies.
