We conducted a retrospective review of patients treated at our thoracic surgery center between January 2013 and August 2024. Inclusion criteria required: (1) age ≥18 years; (2) ECOG performance status 0-1; (3) completely resected stage IB (T2aN0M0) SqCC according to UICC/AJCC 9^th edition staging criteria (3); and (4) completion of ≥2 cycles of platinum-based doublet chemotherapy (without immunotherapy) initiated 4–8 weeks postoperatively. Eligible surgical procedures included lobectomy, sleeve lobectomy, bilobectomy, or pneumonectomy with systematic lymph node dissection. Exclusion criteria comprised: (1) preinvasive or minimally invasive SqCC; (2) metastatic disease; (3) receipt of neoadjuvant therapy; (4) perioperative mortality (death within 30 days post-surgery); (5) protocol non-compliance; or (6) loss to follow-up.

From an initial pool of 1130 screened patients, 181 met all eligibility criteria ( Figure 1 ). Clinicopathological characteristics and survival data were extracted from our institutional electronic medical records system. Adjuvant chemotherapy regimens were selected at the investigator’s discretion from among four standard platinum-doublet options, administered in 21-day cycles.

Adjuvant platinum-based chemotherapy was recommended for stage IB patients presenting with one or more high-risk features, including: poor tumor differentiation, visceral pleural invasion (VPI), sublobar resection, undetermined lymph node status (pNx), lymphovascular invasion (LVI), or spread through air spaces (STAS). Treatment allocation (ACT or observation) was determined through shared decision-making, considering each patient’s performance status (ECOG PS) and personal preferences. This study was conducted in compliance with Good Clinical Practice guidelines and the Declaration of Helsinki. The protocol was approved by the Ethics Committee of Chinese People’s Liberation Army General Hospital (Approval No. S2021-701-01). Written informed consent was obtained from all participants prior to study enrollment.

The follow-up protocol was conducted in accordance with the standardized procedures established in our prior investigation (17). Patient follow-up data were systematically collected through formal channels, including telephone interviews and comprehensive review of outpatient medical records, with the final follow-up conducted in May 2025. The primary study endpoint, DFS, was calculated as the duration (in months) from the surgical intervention to the occurrence of the first event (encompassing tumor recurrence, distant metastasis, or NSCLC-associated mortality) or the last documented follow-up for event-free patients. The secondary endpoint, OS, was defined as the time interval (measured in months) from the date of surgery to death from any cause.

To address potential selection bias resulting from baseline disparities between patients receiving ACT and those under clinical observation (CO), we employed propensity score matching (PSM) with careful consideration of clinically meaningful covariates (18). The matching protocol accounted for seven critical prognostic factors: patient demographics (age, sex), tumor morphology (maximum pathologic diameter), histologic grade (poor differentiation status), and three invasive pathological features (VPI, LVI, and STAS). Using a nearest-neighbor approach with a restricted caliper width (0.02 SD of the propensity score logit) and non-replacement sampling, we established balanced cohorts through 1:1 matching, thereby enhancing the reliability of subsequent comparative analyses while maintaining adequate statistical power.

Statistical analyses were conducted using SPSS 22.0 (IBM Corp., Armonk, NY) with rigorous attention to methodological appropriateness. Continuous variables were characterized using both parametric (mean ± standard deviation) and non-parametric (median, interquartile range) descriptors to accommodate different data distributions. For comparative analyses of continuous measures, the Mann-Whitney U test was employed to account for potential non-normal distributions. Categorical data were presented as absolute counts with corresponding percentages (n, %), with between-group comparisons performed using Pearson’s chi-square test with Yates’ continuity correction where appropriate. Time-to-event outcomes (DFS and OS) were analyzed through Kaplan-Meier survival estimation, with between-strata differences assessed via Mantel-Cox log-rank testing. All probability values were interpreted using a two-tailed α-level of 0.05 as the threshold for statistical significance.

Our institutional database identified 1130 consecutive stage IB NSCLC patients who underwent curative resection between January 2013 and August 2024. After applying stringent inclusion criteria, 181 patients formed the final study cohort ( Table 1 ). The cohort comprised 50 patients (27.6%) who received ACT and 131 patients (72.4%) under CO. Pre-matching analysis revealed significant intergroup differences in several baseline characteristics. Patients in the ACT group demonstrated a younger age distribution (mean 59.6 ± 7.6 vs 63.8 ± 8.1 years, P = 0.002), with a higher proportion under 65 years (72.0% vs 49.6%, P = 0.007). Tumor morphology also differed substantially, with ACT cases presenting larger primary lesions (36.4 ± 3.7 mm vs 31.8 ± 7.3 mm, P < 0.001) and greater prevalence of tumors exceeding 3 cm (92.0% vs 61.8%, P < 0.001). Histopathological assessment showed poorer differentiation patterns more frequently in the ACT cohort (58.0% vs 35.9%, P = 0.017). These pretreatment imbalances were effectively mitigated through PSM, achieving balanced cohorts for subsequent comparative analyses.

The ACT cohort demonstrated distinct preferences in chemotherapy regimen selection, with cisplatin-docetaxel emerging as the most frequently prescribed combination (56%, n=28) ( Table 2 ). Alternative regimens included cisplatin-gemcitabine (16%, n=8) and carboplatin-paclitaxel (28%, n=14). Notably, regimen completion rates varied substantially, with cisplatin-docetaxel achieving perfect adherence (100% completion) compared to 78.6% for carboplatin-paclitaxel. Treatment-related adverse events were common, affecting 80% (n=40) of ACT recipients. Toxicity profiles differed meaningfully across regimens: Cisplatin-gemcitabine demonstrated optimal tolerability with no grade ≥3 toxicities or treatment discontinuations, Carboplatin-paclitaxel showed intermediate toxicity but higher regimen incompletion rates, and Cisplatin-docetaxel, while having higher grade ≥3 events, maintained superior treatment completion. This pattern suggests that while platinum-doublet regimens exhibit class-effect toxicities, specific combinations may offer different risk-benefit profiles in the adjuvant setting for SqCC patients.

The median follow-up time was 60 months (range: 5–137 months). In the entire cohort, twenty-eight patients died and 27 patients had recurrence or metastasis at the last follow-up. Seven patients had recurrence or metastasis in the ACT group, three of them had local recurrence without distant metastasis, and four of them experienced both local recurrence and metastasis. Five of them died even with subsequent therapy. Twenty patients had recurrence or metastasis in the CO group, eight of them had local recurrence without distant metastasis, two of them only found out distant metastasis, and 10 patients experienced both local recurrence and metastasis. Thirteen patients died due to lung cancer-related cause. In the CO group, six patient died of myocardial infarction, one patient died of COVID-19, one patient died of cerebrovascular disease, and one patient died of multiple myeloma. Among the matched 92 patients, 7 (15.2%) patients in the ACT group and 8 (17.4%) patients in the CO group had disease relapse. The 3-year DFS rates were 93.8% in the ACT group and 87.3% in the CO group (HR: 3.12, 95% CI 0.72-13.56; P = 0.130). The 5-year DFS rates were 84.8% in the ACT group and 85.6% in the CO group (HR: 1.05, 95% CI 0.44-2.51; P = 0.919). There was no statistically significant difference in DFS between groups, both in the entire cohort (P = 0.701. Figure 2A ) and the matched cohort (P = 0.665. Figure 2B ). Five (10.9%) patients in the ACT group and 8 (17.4%) patients in the CO group died. The 5-year OS rates were 96.8% in the ACT group and 93.1% in the CO group (HR: 2.46, 95% CI 0.72-8.35; P = 0.150). Meanwhile, no statistically significant difference in OS was observed, either in the overall cohort (P = 0.131. Figure 2C ) or in the matched cohort (P = 0.300. Figure 2D ) between the two groups.

Univariate and subsequent multivariate analyses revealed that age, sex, tumor differentiation, tumor diameter, STAS, LVI, VPI, number of lymph node stations sampled, and ACT were not significantly associated with DFS ( Table 3 ). Similarly, these factors did not demonstrate significant associations with OS in either univariate or multivariate analyses ( Table 4 ).