PURPOSE: Treatment after failure of circumferential resection margin (CRM) conversion after preoperative chemoradiotherapy (pCRT) for locally advanced rectal cancer (LARC) has not been evaluated well. We conducted a single‐center, retrospective analysis to fill this information gap. METHODS: From 2008 to 2016, we included 112 patients who had predictive CRM involvement on baseline magnetic resonance imaging (MRI) and who underwent surgery following pCRT for LARC. Baseline and posttreatment radiologic and clinical factors were analyzed. RESULTS: Of 493 patients with LARC, 112 had CRM involvement by baseline MRI (mrCRM). In 40 patients (35.7%), mrCRM involvement was converted as negative posttreatment CRM (ymrCRM−). Multivariate analysis showed the risk factors for persistent CRM involvement (ymrCRM+) after pCRT were extramural venous invasion (mrEMVI+) (P = 0.030) and lower tumor location (P = 0.007). In addition, persistent CRM involvement after pCRT was an independent risk factor for predicting pathologic CRM involvement. The Cox proportional hazard model showed baseline positive mrEMVI remained significant for disease-free survival (DFS) (P < 0.001). On posttreatment MRI, abdominoperineal resection (P = 0.031), intersphincteric resection (P = 0.006), and persistent CRM involvement (P = 0.001) remained significant for local recurrence-free survival. With regard to DFS, persistent CRM involvement (P = 0.048) and positive EMVI on posttreatment MRI (ymrEMVI) (P = 0.014) were significant. In the patient subgroup with persistent CRM involvement, 5-year DFS in patients with mrEMVI and ymrEMVI was 29.8% and 21.2%, respectively. CONCLUSION: Patients who fail to convert to negative CRM have extremely poor oncologic outcomes. Lower tumor height and negative mrEMVI status were good responders to ymrCRM conversion. Our results suggest that these patients require a more intensive treatment modality.
Chemoradiotherapy ; Disease-Free Survival ; Humans ; Magnetic Resonance Imaging ; Multivariate Analysis ; Proportional Hazards Models ; Rectal Neoplasms ; Retrospective Studies ; Risk Factors