Related Experiment Video
Updated: Aug 17, 2026
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
Cumulative high-risk surgical and pathologic features predict recurrence after resection of early-stage node-negative
Fatemehsadat Pezeshkian1, Emanuele Mazzola2, Justin S Heidel1
1Division of Thoracic Surgery, Mass General Brigham, Boston, Mass.
Objective:
Despite curative resection, recurrence remains a clinically relevant problem in patients with early-stage non-small cell lung cancer (NSCLC). Although several high-risk features have been individually associated with adverse outcomes, their cumulative prognostic influence remains unclear.
Methods:
Patients with NSCLC undergoing curative-intent resection between 2017 and 2024 were identified from a prospective institutional database. Inclusion criteria were solid tumors with size <4 cm, pN0 disease, and R0 resection with documented spread through air spaces status. High-risk features included nonanatomic resection, poorly/undifferentiated grade, lymphovascular invasion (LVI), visceral pleural invasion (VPI), incomplete lymphadenectomy (<3N2 and <1N1 stations), and spread through air spaces positivity. Cox proportional hazards modeling was used to quantify the association of high-risk features and 5-year recurrence-free survival, and a nomogram was constructed to estimate cumulative recurrence risk.
Results:
Of the 497 patients included, anatomic resection was performed in 268 (53.9%) patients and complete lymphadenectomy was achieved in 126 (25.4%). Over a median 30-month follow-up (interquartile range, 13-49 months), 50 patients (10.1%) developed recurrence. LVI (hazard ratio [HR], 3.7; 95% CI, 2-7; P < .001), VPI (HR, 2.8; 95% CI, 1.5-5.2; P = .002), poorly/undifferentiated tumor (HR, 1.8; 95% CI, 0.99-3.3; P = .05), resection margin distance (HR, 0.7; 95% CI, 0.5-0.9; P = .01), and incomplete lymphadenectomy (HR, 2.7; 95% CI, 1-7.1; P = .05) were significantly associated with recurrence. The final nomogram demonstrated good discrimination for estimated 5-year recurrence risk, with a concordance index of 0.82.
Conclusions:
A multifactorial nomogram integrating high-risk pathologic features to estimate recurrence risk in early-stage NSCLC was developed. This model suggests that recurrence risk in early-stage NSCLC may be driven by the cumulative burden of high-risk surgical and pathologic features rather than isolated factors, providing a clinically applicable framework for postoperative risk stratification. External validation and integration into clinical decision-making frameworks are warranted.