Development and Validation of an Elastic Net– Softmax Machine Learning Model for Predicting Lymph Node Regression Grade After Neoadjuvant Chemoradiotherapy in Locally Advanced Rectal Cancer

Abstract Background: Neoadjuvant chemoradiotherapy (nCRT) followed by total mesorectal excision is standard for locally advanced rectal cancer (LARC). While tumor regression grade is routinely reported, lymph node regression grade (LRG) is underutilized. We aimed to develop a machine learning model for preoperative LRG prediction. Methods: We retrospectively reviewed 111 LARC patients who underwent nCRT plus surgery between January 2016 and December 2020. LRGmax (grades 0–5) was assessed using Mandard criteria. Elastic Net regression with 10-fold cross-validation was used for feature selection, followed by a Softmax multinomial classifier. Patients were randomly split into training (70%) and validation (30%) sets. Performance was evaluated using AUC, accuracy, sensitivity, specificity, precision, and recall. Bootstrap resampling and decision curve analysis assessed calibration and clinical net benefit. Results: The LRGmax distribution was: grade 0 (28.8%), 1 (39.6%), 2 (12.6%), 3 (10.8%), 4 (5.4%), and 5 (2.7%). In the training set, AUC was 0.99. In the validation set, AUC was 0.86 and specificity 0.87. Calibration curves showed high consistency; DCA revealed positive net benefit. Higher LRGmax correlated with increased nodal metastasis count, lymph node ratio, and worse survival (all P < 0.001). LRGmax correlated positively with radiotherapy efficacy ( r = 0.28, P = 0.003) and negatively with 3-year survival ( r = –0.47, P < 0.001). Conclusions: The Elastic Net–Softmax model showed discriminative utility for predicting LRGmax after nCRT in LARC. Despite decreased validation metrics due to limited sample size and class imbalance, the model retained predictive value as a proof-of-concept tool. Multi-center prospective studies are warranted.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-13
DOI
https://doi.org/10.21203/rs.3.rs-10882836/v1
Primary Topic
Colorectal Cancer Surgical Treatments
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Development and Validation of an Elastic Net– Softmax Machine Learning Model for Predicting Lymph Node Regression Grade After Neoadjuvant Chemoradiotherapy in Locally Advanced Rectal Cancer

Ran Zhang, Shuyuan ZHAO, Yan Liu, Kai Liu
4 links
Research Square
Colorectal Cancer Surgical Treatments
preprint

Development and Validation of an Elastic Net– Softmax Machine Learning Model for Predicting Lymph Node Regression Grade After Neoadjuvant Chemoradiotherapy in Locally Advanced Rectal Cancer

Ran Zhang, Shuyuan ZHAO, Yan Liu, Kai Liu
preprint en

Abstract

Abstract Background: Neoadjuvant chemoradiotherapy (nCRT) followed by total mesorectal excision is standard for locally advanced rectal cancer (LARC). While tumor regression grade is routinely reported, lymph node regression grade (LRG) is underutilized. We aimed to develop a machine learning model for preoperative LRG prediction. Methods: We retrospectively reviewed 111 LARC patients who underwent nCRT plus surgery between January 2016 and December 2020. LRGmax (grades 0–5) was assessed using Mandard criteria. Elastic Net regression with 10-fold cross-validation was used for feature selection, followed by a Softmax multinomial classifier. Patients were randomly split into training (70%) and validation (30%) sets. Performance was evaluated using AUC, accuracy, sensitivity, specificity, precision, and recall. Bootstrap resampling and decision curve analysis assessed calibration and clinical net benefit. Results: The LRGmax distribution was: grade 0 (28.8%), 1 (39.6%), 2 (12.6%), 3 (10.8%), 4 (5.4%), and 5 (2.7%). In the training set, AUC was 0.99. In the validation set, AUC was 0.86 and specificity 0.87. Calibration curves showed high consistency; DCA revealed positive net benefit. Higher LRGmax correlated with increased nodal metastasis count, lymph node ratio, and worse survival (all P < 0.001). LRGmax correlated positively with radiotherapy efficacy ( r = 0.28, P = 0.003) and negatively with 3-year survival ( r = –0.47, P < 0.001). Conclusions: The Elastic Net–Softmax model showed discriminative utility for predicting LRGmax after nCRT in LARC. Despite decreased validation metrics due to limited sample size and class imbalance, the model retained predictive value as a proof-of-concept tool. Multi-center prospective studies are warranted.

Research Square
Xinjiang Medical University (CN)
Xinjiang Medical University
Good health and well-being
Colorectal Cancer Surgical Treatments
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon