Predictive value of circulating lymphocyte subsets and inflammatory indexes for neoadjuvant chemoradiotherapy response in rectal mucinous adenocarcinoma patients: A machine learning approach

Yu Lin ¹, Yanwu Sun ¹, Weizhong Jiang ¹

⁰Department of Colorectal Surgery, Fujian Medical University Union Hospital, Fuzhou, Fujian, People's Republic of China.

Cancer Medicine +

|

July 24, 2024

View abstract on PubMed

Summary

This study shows that a machine learning model using inflammation markers and lymphocyte subsets can predict treatment response in rectal mucinous adenocarcinoma patients. This offers a new tool for personalized therapy in rectal cancer.

Area Of Science

Oncology
Immunology
Medical Informatics

Background

Rectal mucinous adenocarcinomas (MACs) present unique challenges in treatment response prediction.
Neoadjuvant chemoradiotherapy (NCRT) is a standard treatment, but predicting patient response remains critical.
Identifying reliable biomarkers for treatment response is essential for optimizing patient outcomes.

Purpose Of The Study

To evaluate the predictive value of circulating lymphocyte subsets and inflammatory indexes for NCRT response in rectal MAC patients.
To develop and compare predictive models (nomogram and machine learning) for treatment response.
To identify key factors influencing radiochemotherapy sensitivity.

Main Methods

Retrospective analysis of 283 rectal MAC patients treated with NCRT and curative resection.
Development of a nomogram using multivariate logistic regression.
Construction of a machine learning (ML) model utilizing the extreme gradient boosting (XGB) algorithm.
Quantification of feature importance using Shapley additive explanations (SHAP).

Main Results

The ML model achieved a higher prediction accuracy (AUROC 0.824 in training, 0.762 in tuning set) compared to the nomogram (C-index 0.756).
Key predictors identified by the ML model included Th/Tc ratio, neutrophil-to-lymphocyte ratio, Th lymphocytes, Gross type, and T lymphocytes.
Tumor length, pretreatment clinical T stage, and PNI were also identified as independent risk factors.

Conclusions

Systemic inflammation and lymphocyte-mediated immune responses significantly influence radiochemotherapy sensitivity in rectal MAC.
The developed ML model, integrating clinical data, lymphocyte subsets, and inflammatory indexes, shows potential as an assessment tool.
This model can aid in providing a reference for individualized treatment strategies for rectal MAC patients.

Keywords:

chemoradiotherapy circulating lymphocyte inflammatory indexes machine learning mucinous adenocarcinoma rectal cancer