GS03 - Lost in Staging: A Retrospective Cohort Study of MRI and Pathologic Staging Concordance in Rectal Cancer

Image or Table (Do Not add text here):

Purpose/Background: High-resolution pelvic MRI is the cornerstone of rectal cancer staging, offering insights into tumor depth, nodal involvement, and risk factors such as extramural vascular invasion (EMVI) and circumferential resection margin (CRM) proximity. However, discordance between MRI and final pathologic staging remains a major clinical challenge, especially in post-neoadjuvant chemoradiotherapy (nCRT) and can result in over or under treatment. This study evaluates the patterns of MRI-pathologic discordance, identifies predictors of inaccuracy, and presents a novel analysis of discordance trends.

Methods/Interventions: A retrospective review was performed on 174 rectal cancer patients who underwent pre-treatment MRI and definitive surgical resection between 2016–2023. MRI and pathologic stages were categorized as Stage 0–IV and compared for concordance. Logistic regression assessed predictors of mismatch, including EMVI, CRM involvement, mesorectal fascia (MRF) involvement, anal sphincter invasion, perineural invasion, molecular subtype, and treatment type (neoadjuvant vs surgery only). Concordance rates were visualized using heatmaps and forest plots.

Results/Outcomes: MRI-pathology concordance was 33.3%; 55.7% were overstaged and 10.9% understaged. Neoadjuvant therapy independently predicted discordance (OR 3.15, p< 0.001); EMVI trended toward significance (OR 6.75, p=0.074); other factors were not predictive. MRI performed better in surgery-only patients. Heatmaps revealed non-random clustering: overstaging was most frequent in MRI Stage II, while understaging clustered in MRI Stage I and IV. A novel observation noted a considerable subset of tumors staged as Stage I by MRI were found to be Stage II or III on pathology. This unique finding reflects MRI’s limited sensitivity for subtle perirectal fat invasion or micrometastatic nodal disease. These patients, assumed low-risk on MRI, may be inappropriately excluded from neoadjuvant therapy. This pattern contrasts with prior emphasis on MRI overstaging and adds a new dimension to the staging debate.

Conclusion/Discussion: MRI staging of rectal cancer shows frequent discordance with pathology, particularly after neoadjuvant therapy. Beyond confirming known trends of overstaging, our study identifies a previously underrecognized pattern of understaging in MRI Stage I tumors with important clinical implications. This directional bias, revealed by heatmap analysis, highlights MRI’s limitations for early T3 or nodal disease and underscores the need for multidisciplinary interpretation of low-risk MRI findings. Tailored staging strategies incorporating radiomics or AI-assisted imaging may help close this accuracy gap and optimize treatment selection for rectal cancer patients.