Objectives To compare MRI and contrast-enhanced CT in assessing neoadjuvant therapy (NAT) response for locally advanced esophageal squamous cell carcinoma (ESCC). Materials and methods In this prospective single-center study (July 2019–August 2024), 122 ESCC patients (cT2-T4a/N0–N3/M0; mean age 63.2 ± 7.0 years; 95 male) underwent pre-/post-NAT MRI and contrast-enhanced CT before esophagectomy. Two blinded radiologists independently evaluated treatment response on CT and MRI images using Response Evaluation Criteria in Solid Tumors (RECIST) and MR tumor regression grade (mrTRG). Postoperative pathologic TRG (pTRG) served as the reference standard. Categories included pathological complete response (pCR: pTRG0, n = 32) vs non-pCR (pTRG1–3, n = 90), and pathological good response (pGR: pTRG0–1, n = 48) vs non-pGR (pTRG2–3, n = 74). Imaging assessments were compared with pathological findings using χ2 test and Cohen’s kappa value. Results CT demonstrated accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of 63.9%, 25.0%, 77.8%, 28.6%, and 77.8%, respectively, in differentiating CR from non-CR, while MRI showed superior performance with respective values of 93.4%, 90.6%, 94.4%, 85.3%, and 96.6%. Cohen’s kappa coefficient revealed high concordance between MRI findings and postoperative pathology in CR evaluation (κ = 0.834, p < 0.001), whereas CT showed no statistically significant agreement (κ = 0.029, p = 0.748). mrTRG had excellent correlation with pTRG (κ = 0.917 for CR/non-CR; κ = 0.710 for GR/non-GR, p < 0.001). Conclusions MRI outperformed CT in post-NAT response evaluation using RECIST for locally advanced ESCC. The mrTRG system enables precise TRG stratification, suggesting MRI as the preferred imaging modality for RECIST implementation in ESCC management. Key Points <Emphasis Type="BoldItalic">Question</Emphasis> Can RECIST be applied to CT or MRI, accurately assess NAT response in locally advanced ESCC, and which modality demonstrates superior performance? <Emphasis Type="BoldItalic">Findings</Emphasis> MRI surpassed CT in discriminating complete response (CR) from non-CR using RECIST and provided a mrTRG system. <Emphasis Type="BoldItalic">Clinical relevance</Emphasis> MRI-based RECIST evaluation, combined with the mrTRG, enables precise risk stratification and personalized treatment planning for locally advanced ESCC, with better clinical utility and pathological correlation compared to CT-based measurements. Graphical Abstract

MR outperformed CT in evaluating RECIST response to neoadjuvant therapy in locally advanced esophageal squamous cell carcinoma

期刊论文
发表日期：

2025年11月18日

查看全文

Wenshi Li,
Zhaoqi Wang,
Zihan Wang,
Yafeng Dong,
Funing Chu,
Yue Huang,
Mengzhu Wang,
Omar Darwish,
Dominik Nickel,
Keke Zhao,
Yan Zheng,
Ihab R. Kamel,
Jinrong Qu

摘要

Objectives

To compare MRI and contrast-enhanced CT in assessing neoadjuvant therapy (NAT) response for locally advanced esophageal squamous cell carcinoma (ESCC).

Materials and methods

In this prospective single-center study (July 2019–August 2024), 122 ESCC patients (cT2-T4a/N0–N3/M0; mean age 63.2 ± 7.0 years; 95 male) underwent pre-/post-NAT MRI and contrast-enhanced CT before esophagectomy. Two blinded radiologists independently evaluated treatment response on CT and MRI images using Response Evaluation Criteria in Solid Tumors (RECIST) and MR tumor regression grade (mrTRG). Postoperative pathologic TRG (pTRG) served as the reference standard. Categories included pathological complete response (pCR: pTRG0, n = 32) vs non-pCR (pTRG1–3, n = 90), and pathological good response (pGR: pTRG0–1, n = 48) vs non-pGR (pTRG2–3, n = 74). Imaging assessments were compared with pathological findings using χ² test and Cohen’s kappa value.

Results

CT demonstrated accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of 63.9%, 25.0%, 77.8%, 28.6%, and 77.8%, respectively, in differentiating CR from non-CR, while MRI showed superior performance with respective values of 93.4%, 90.6%, 94.4%, 85.3%, and 96.6%. Cohen’s kappa coefficient revealed high concordance between MRI findings and postoperative pathology in CR evaluation (κ = 0.834, p < 0.001), whereas CT showed no statistically significant agreement (κ = 0.029, p = 0.748). mrTRG had excellent correlation with pTRG (κ = 0.917 for CR/non-CR; κ = 0.710 for GR/non-GR, p < 0.001).

Conclusions

MRI outperformed CT in post-NAT response evaluation using RECIST for locally advanced ESCC. The mrTRG system enables precise TRG stratification, suggesting MRI as the preferred imaging modality for RECIST implementation in ESCC management.

Key Points

Question Can RECIST be applied to CT or MRI, accurately assess NAT response in locally advanced ESCC, and which modality demonstrates superior performance?

Findings MRI surpassed CT in discriminating complete response (CR) from non-CR using RECIST and provided a mrTRG system.

Clinical relevance MRI-based RECIST evaluation, combined with the mrTRG, enables precise risk stratification and personalized treatment planning for locally advanced ESCC, with better clinical utility and pathological correlation compared to CT-based measurements.

Graphical Abstract