Dynamic radiological features predict pathological response after neoadjuvant immunochemotherapy in esophageal squamous cell carcinoma

Yuli Ruan, Yue Ma, Ming Ma, Chao Liu, Dan Su, Xin Guan, Rui Yang, Hong Wang, Tianqin Li, Yang Zhou, Jianqun Ma, Yanqiao ZhangJournal of Translational Medicine 2024 · DOI

TL;DR

This study introduces a novel imaging score (ECPI-Score) derived from dynamic CT scans to predict pathological complete response (pCR) in patients with locally advanced esophageal squamous cell carcinoma (LA-ESCC) undergoing neoadjuvant immunochemotherapy (NICT). The ECPI-Score, combined with the presence or absence of a vascular sign, shows high accuracy in predicting treatment response, offering clinicians a valuable tool for personalized treatment planning.

What Problem Does This Paper Address?

Locally advanced esophageal squamous cell carcinoma (LA-ESCC) is a serious condition, and neoadjuvant immunochemotherapy (NICT) followed by surgery has emerged as a promising treatment. A key indicator of successful treatment is pathological complete response (pCR), meaning no residual cancer cells are found in the surgically removed tissue. However, reliably predicting pCR before treatment is challenging. While biomarkers like PD-L1 and tumor mutation burden exist, they don't always correlate well with pCR. This leaves a critical gap in identifying patients who will benefit most from NICT, potentially leading to suboptimal treatment strategies. The paper aims to address this by exploring dynamic radiological features from CT scans as predictive markers for pCR.

Methods

The researchers developed a novel scoring system, the ESCC preoperative imaging score (ECPI-Score), by integrating four key dynamic CT imaging features: post-treatment maximal CT value, percentage change in tumor thickness, post-treatment maximal tumor thickness, and percentage change in maximal tumor length. These features were selected based on their ability to predict pCR. The study also investigated the predictive value of the "vascular sign" (a qualitative assessment of vascularity on post-treatment CT scans).

Logistic regression models were used to assess the independent predictive power of the ECPI-Score and vascular sign for pCR. The performance of these predictors was evaluated using the area under the receiver operating characteristic curve (AUC). The study also employed Kaplan-Meier analysis to assess the impact of these features on recurrence-free survival (RFS) and overall survival (OS).

Datasets

The study retrospectively analyzed data from 192 patients with LA-ESCC who underwent surgery following NICT at Harbin Medical University Cancer Hospital between December 2019 and October 2023. The dataset included pretreatment and preoperative CT images, as well as pathological findings. No new benchmark datasets were introduced.

Key Insights and Ideas

  • Dynamic CT is Powerful: The study highlights that changes in tumor characteristics observed on CT scans during and after neoadjuvant therapy are highly informative for predicting treatment outcomes. This moves beyond static imaging assessments.
  • ECPI-Score as a Composite Predictor: Combining multiple dynamic CT features into a single score (ECPI-Score) significantly improves the prediction of pCR compared to individual features. This suggests a synergistic effect of these radiological markers.
  • Vascular Sign as a Simple, Yet Effective, Indicator: The presence or absence of a vascular sign on post-treatment CT scans emerged as an independent predictor of pCR, demonstrating that even qualitative imaging assessments can be clinically valuable.
  • Predicting More Than Just Response: While the primary focus is pCR, the study also suggests that these radiological features (ECPI-Score and vascular sign) may correlate with improved recurrence-free survival, hinting at their broader prognostic value.

Key Results

  • The developed ECPI-Score demonstrated a high predictive capability for pCR, with an AUC of 0.918 in the training set and 0.862 in the validation set.
  • Both the ECPI-Score and the vascular sign were identified as independent predictors of pCR in multivariate analysis.
  • Patients with a high ECPI-Score (≥2) had a significantly higher pCR rate (62.07%) compared to those with a low ECPI-Score (<2) (5.97%).
  • Similarly, patients with a negative vascular sign after treatment had a higher pCR rate (72.0%) compared to those with a positive vascular sign (28.00%).
  • The study found that patients achieving pCR had significantly better recurrence-free survival (RFS) (P=0.036) compared to those with non-pCR.
  • A negative vascular sign was also associated with significantly better RFS (P=0.038).

Limitations and Future Work

The authors acknowledge several limitations:

  • Retrospective Nature: The study design is retrospective, which can introduce biases.
  • Single-Center Study: The findings are based on data from one institution, limiting generalizability.
  • Subjectivity of Vascular Sign: The assessment of the vascular sign is described as qualitative, which can be subject to inter-observer variability.

For future work, the authors suggest:

  • Multi-center Validation: Conducting studies across multiple centers to validate the findings and confirm the generalizability of the ECPI-Score and vascular sign.
  • Artificial Intelligence Exploration: Investigating the potential of artificial intelligence (AI) and machine learning to overcome limitations related to imaging resolution and observer subjectivity, potentially leading to more objective and robust predictive models.
  • Tailored Treatment Strategies: Further exploring how these predictive markers can inform tailored treatment approaches, especially for patients with tumors in challenging locations or those who do not respond to NICT.

Summary updated

Abstract

This study demonstrates dynamic radiological features are independent predictors of pCR for LA-ESCC treated with NICT. It will guide clinicians to make accurate treatment plans.
BibTeX
@article{ruan2024dynamic,
  author = {Yuli Ruan and Yue Ma and Ming Ma and Chao Liu and Dan Su and Xin Guan and Rui Yang and Hong Wang and Tianqin Li and Yang Zhou and Jianqun Ma and Yanqiao Zhang},
  title = {{Dynamic radiological features predict pathological response after neoadjuvant immunochemotherapy in esophageal squamous cell carcinoma}},
  journal = {Journal of Translational Medicine},
  year = {2024},
  doi = {10.1186/s12967-024-05291-8},
}
Figure