Updated perspectives on visceral pleural invasion in non-small cell lung cancer: A propensity score-matched analysis of the SEER database

Curr Probl Cancer. 2025 Apr 18;56:101205. doi: 10.1016/j.currproblcancer.2025.101205. Online ahead of print.

ABSTRACT

BACKGROUND: Visceral pleural invasion (VPI), including PL1 (the tumor invades beyond the elastic layer) and PL2 (the tumor extends to the surface of the visceral pleura), plays a crucial role in staging Non-Small Cell Lung Cancer (NSCLC). However, there is a growing debate concerning the prognostic significance of PL1 and PL2. This study, therefore, conducted the analysis of the prognostic differences between PL1 and PL2 to inform more precise staging and treatment strategies.

METHODS: Altogether, 12,223 resected T1-3N0M0 NSCLC patients from 2010 to 2015 were enrolled. Utilizing propensity score matching (PSM) and Kaplan-Meier survival analysis, this study explored the prognosis of patients under different settings of VPI and the impact of various treatments. Finally, a machine learning model was constructed to accurately predict the 5-year survival probability.

RESULTS: For tumors ≤ 50 mm, PL1 did not confer a survival disadvantage compared to PL0 (the tumor within the elastic layer of the visceral pleura), whereas PL2 did. Notably, patients with tumor sizes 31-50 mm and PL2 have a similar poor prognosis to patients with tumor sizes of 51-70 mm and PL0. Further survival analysis showed that lobectomy offered better outcomes than sublobectomy. Moreover, patients in this study did not benefit from postoperative radiotherapy or chemotherapy. A model with high efficacy in predicting the 5-year survival probability was developed eventually.

CONCLUSION: These data support the viewpoint that staging patients with tumor ≤ 30 mm and PL1 as T1. Those with 31-50 mm tumors and PL2, exhibiting a similar poor prognosis to patients with T3 and PL0, warrant a T3 classification. Apart from optimizing the TNM staging system, machine learning could also play a significant role in prognostic prediction.

PMID:40252301 | DOI:10.1016/j.currproblcancer.2025.101205