A Preoperative Risk Scoring System for Survival Prediction in Clinical Stage IB Lung Adenocarcinoma: A Multicenter Study

Kotaro Murakami^1,2*Tetsuya Isaka¹Takuya Nagashima¹Hiroyuki Adachi¹Shunsuke Shigefuku¹Noritake Kikunishi¹Naoko Shigeta¹Yujin Kudo²Yoshihiro Miyata³Morihito Okada³Norihiko Ikeda²Hiroyuki Ito¹

• ¹Department of Thoracic Surgery, Kanagawa Cancer Center, Yokohama, Japan
• ²Department of Surgery, Tokyo Medical University, Tokyo, Japan
• ³Department of Surgical Oncology, Hiroshima University, Hiroshima, Japan

Background: Clinical stage (c-stage) IB lung adenocarcinoma (LUAD) presents variable survival outcomes, and the prognostic significance of factors such as ground-glass opacity components and positron emission tomography (PET) metrics remains unclear. Despite recent advances, no preoperative scoring model has been established to stratify risk in this subgroup. We aimed to identify preoperative prognostic factors in c-stage IB LUAD and develop a simple scoring system for predicting overall survival (OS). Methods: We retrospectively analyzed data from 245 patients with c-stage IB LUAD who underwent lobectomy at three institutions between 2010 and 2020. Cox regression analysis was performed to identify independent preoperative prognostic factors for OS. A risk score was developed by assigning points to each factor, based on the regression coefficients. Thereafter, patients were stratified into four risk groups based on their total scores. Results: Multivariate analysis identified smoking history (hazard ratio [HR]: 2.68; 95% confidence interval [CI]: 1.13–6.33; p=0.025), elevated serum carcinoembryonic antigen (CEA) levels (HR: 2.89; 95%CI: 1.42– 5.91; p=0.004), and high maximum standardized uptake value (SUVmax) on PET (HR: 2.84; 95%CI: 1.16– 6.98; p=0.023) as independent factors of poor prognosis. A scoring system was established by assigning one point to each factor. Patients were stratified into four risk groups: low (score 0, n=41), moderate (score 1, n=84), moderately high (score 2, n=77), and extremely high (score 3, n=43). Five-year OS rates were 100.0%, 89.3%, 74.0%, and 52.1%, respectively (p<0.001). The prognostic model demonstrated acceptable discrimination ability (area under the curve [AUC], 0.738; 95%CI, 0.661–0.815) and concordance index (AUC, 0.753; 95%CI, 0.682–0.824). Notably, patients with a score of 0 showed low-grade tumors and favorable prognosis, whereas those with a score of 3 had more aggressive pathological characteristics and significantly worse outcomes. Conclusions: We developed and validated a simple preoperative scoring system using smoking history, serum CEA level, and tumor SUVmax to predict prognosis in c-stage IB LUAD. This model provides a practical tool for risk stratification and may support individualized treatment decisions, including the consideration of induction therapy in selected cases.