Value of MRI in Predicting Pathologic Complete Response after Neoadjuvant Chemoradiation with or without Chemotherapy (NACRT) for Locally Advanced Rectal Cancer (LARC).

Abstract

<h3>Purpose/Objective(s)</h3> LARC patients with a complete response after NACRT are eligible for organ preservation obviating the need for surgery. MRI evaluation for radiologic complete response (rCR) is essential in determining who is a candidate for organ preservation (OP). We evaluate the ability of MRI, after NACRT, to predict pathologic complete response (pCR). <h3>Materials/Methods</h3> Patients with LARC (T3-4, N0-2) with post NACRT MRI from 2013 to 2021 were retrospectively identified. MRI and pathology reports were queried for treatment response. Positive (PPV) and negative (NPV) predictive values were calculated and used to test the discriminative ability (ROC curve) of MRI post NACRT. Multivariate logistic regression of post-NACRT MRI characteristics, TN staging, lymph node (LN) size, extramural venous invasion, CEA, smoking pack years, RT dose, total neoadjuvant therapy (TNT), sex, and race were used to identify predictors for rCR and pCR. pCR was defined as no evidence of disease after surgery or flexible sigmoidoscopy with biopsy. <h3>Results</h3> Of the 69 patients analyzed, 71% were male and 25% were non-white. The median age was 58 years. Overall pCR rate was 41% whereas OP rate (no more than local excision) was 29%. PPV and NPV of MRI for the primary mass were 72% and 77% respectively with an area under ROC curve of 0.74. Higher pre-NACRT CEA (OR=0.92; CI 0.85-0.98, p=0.019) and higher RT dose (OR=0.99; CI 0.99-0.99, p=0.047) were associated with rCR of the primary mass. Complete fibrosis on MRI was associated with pCR of the primary mass (OR=0.15; CI 0.02-0.99, p=0.049). Of the 11 patients that had diffusion restricted mass after NACRT, all were regarded as non-rCR and 8 were non-pCR at surgery/biopsy. PPV and NPV of MRI for LN were 95% and 42% respectively with an area under ROC of 0.79. TNT (OR=0.05; CI 0.002-0.93, p=0.045), higher CEA (OR=0.92; CI 0.84-0.99, p=0.045) and smoking pack years (OR=0.91; CI 0.83-0.99, p=0.029) were associated with rCR and pCR congruence. However, time between post NACRT MRI and surgery/biopsy was not significantly associated with congruence. <h3>Conclusion</h3> The discriminative ability of MRI to predict pCR of primary rectal mass was middling. Patients with increasing pre-NACRT CEA and received higher RT dose were more likely to achieve rCR. Complete fibrosis on post NACRT MRI was our best predictor of pCR whereas restricted diffusion on MRI correlated with non-pCR. There is room for improvement in OP patient selection by pelvic MRI in the post-NACRT setting.