Nomogram Development for Predicting Synchronous Lung Metastasis in Patients with T1 Colorectal Cancer: An SEER-Based Analysis

TL;DR

This study created two tools to predict lung metastasis in early-stage colorectal cancer patients, helping decide when to use CT scans and improve prognosis.

Contribution

Developed and validated two nomograms for predicting synchronous lung metastasis in T1 colorectal cancer patients using SEER data.

Findings

Model A (clinicopathologic-only) showed moderate discrimination (AUC = 0.728) for pre-imaging risk stratification.

Model B (including concurrent organ metastasis) demonstrated good discrimination (AUC = 0.856) for post-staging risk profiling.

Calibration plots confirmed good agreement between predicted and observed probabilities for both models.

Abstract

Background and Objectives: Colorectal cancer is a significant global health burden, with lung metastasis contributing substantially to mortality. Accurate risk stratification of synchronous lung metastasis (sLM) in patients with T1 colorectal cancer is important for informing staging decisions, yet no validated tool exists to guide selective chest computed tomography (CT) in this population. This study aimed to develop and validate two complementary nomograms: a clinicopathologic model (Model A) for pre-imaging risk stratification to guide chest CT decisions, and a post-staging model (Model B) incorporating concurrent organ metastasis status for comprehensive risk profiling. Materials and Methods: We utilized data from the Surveillance, Epidemiology, and End Results database, including patients diagnosed with T1 colorectal cancer between 2010 and 2020. Logistic regression analyses identified significant predictors of synchronous lung metastasis. A nomogram was constructed based on these predictors and validated using a split-sample approach. Results: The study included 41,728 patients with T1 colorectal cancer. Significant predictors of synchronous lung metastasis included tumor grade, size, location, lymph node involvement, and concurrent metastases in other organs. Two models were developed: Model A (clinicopathologic-only) demonstrated moderate discriminatory ability (AUC = 0.728, 95% CI: 0.710–0.746), while Model B (including concurrent organ metastasis status) demonstrated good discrimination (AUC = 0.856, 95% CI: 0.843–0.869): Model A validation AUC = 0.716; Model B validation AUC = 0.849. Calibration plots showed good agreement between predicted and observed probabilities of synchronous lung metastasis. Conclusions: This study developed and internally validated two nomograms for predicting sLM in patients with T1 CRC. Model A, using readily available clinicopathological factors, may support selective chest CT decisions during initial staging. Model B, incorporating post-staging information, may assist in prognostic counseling. External validation is required before clinical implementation.

Figures