Grade-driven adjuvant chemotherapy benefit in high-risk stage IB NSCLC: a retrospective cohort study.

Introduction
Non-small cell lung cancer (NSCLC) is the most common type of lung cancer and a leading cause of cancer-related death worldwide, with adenocarcinoma as the predominant histologic subtype. In patients with completely resected stage II-IIIA NSCLC, multiple randomized controlled trials (RCTs) have demonstrated that adjuvant chemotherapy (ACT) significantly improves overall survival (OS) and disease-free survival (DFS). However, the benefit of postoperative ACT for patients with completely resected stage IB NSCLC remains controversial (1-7).
The 2025 National Comprehensive Cancer Network (NCCN) guidelines do not routinely recommend ACT for stage IB patients, but suggest that it may be considered for those with high-risk features such as visceral pleural invasion (VPI), lymphovascular invasion (LVI), or poor differentiation (8). Similarly, the American Society of Clinical Oncology (ASCO) advises against routine postoperative ACT in stage IB disease, instead recommending individualized risk-benefit discussions with a medical oncologist (evidence quality: intermediate) (9). The Chinese Society of Clinical Oncology (CSCO) guidelines recommend ACT for stage IB patients with high-risk features (Category 2A evidence), and suggest that it may be considered for tumors with predominantly solid or micropapillary components (Category 2B evidence). In contrast, the European Society for Medical Oncology (ESMO) does not recommend ACT for patients with stage IB disease (10).
Historically, several RCTs have investigated ACT in patients with stage IB–IIIA NSCLC. The CALGB 9633 trial initially reported improved survival with adjuvant carboplatin plus paclitaxel in stage IB patients with tumors ≥4 cm (11). Similarly, the JBR-10 trial found that patients with tumors ≥4 cm might benefit from ACT (12). In contrast, the ANITA found no significant survival benefit in stage IB patients. However, these trials were based on the 7th edition of the Tumor, Node, Metastasis (TNM) staging system, in which tumors measuring 4–5 cm (T2bN0M0) were classified as stage IB (7,13). According to the updated 9th edition of the TNM staging system, tumors ≥4 cm are now classified as stage IIA (14). This reclassification limits the applicability of previous RCTs to current-stage IB patients, especially those with T2aN0M0 tumors.
Histologic subtype is recognized as a key prognostic factor in lung adenocarcinoma. In 2020, the International Association for the Study of Lung Cancer (IASLC) proposed a novel grading system based on the proportion of high-grade components such as solid, micropapillary, and complex glandular patterns (14). This system stratifies tumors into low-, intermediate-, and high-grade groups based on the predominant histologic pattern and the presence of >20% high-grade components (15). Several studies have examined the prognostic value of micropapillary or solid components in stage IB adenocarcinoma, but results remain inconsistent (16-19). Moreover, the clinical utility of the IASLC grading system in guiding ACT decisions for high-risk stage IB patients remains unclear.
Therefore, we conducted this retrospective cohort study to evaluate the efficacy of ACT in stage IB adenocarcinoma patients with high-risk features, with a particular focus on histologic grade as a potential modifier of treatment benefit. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1858/rc).

Methods

Patients
We retrospectively reviewed all patients who underwent complete resection for pathologic stage IB lung adenocarcinoma at Jinling Hospital between January 2016 and December 2019.
The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Institutional Review Board of the Jinling Hospital (approval No. DZQH-KYLL-25-08) and individual consent for this retrospective analysis was waived.
Tumor stage was reassessed manually according to the 9th edition of the American Joint Committee on Cancer (AJCC) TNM classification system [2024], based on each patient’s original pathological report, to ensure staging consistency. Only patients with pathologic stage IB disease (T2aN0M0, tumor diameter >3 and ≤4 cm) were included. The restaging and histopathologic evaluations were performed independently by two board-certified pulmonary pathologists who were blinded to adjuvant treatment information and clinical outcomes. Any discrepancies in tumor staging or histologic interpretation were resolved through consensus review.
Inclusion criteria were as follows:
Patients who underwent complete resection as initial treatment and did not receive neoadjuvant therapy.

Pathologically confirmed stage IB lung adenocarcinoma according to the 9th edition TNM classification.

Presence of at least one high-risk feature: VPI, LVI, or poor tumor differentiation.

Patients who received four cycles of platinum-based ACT following surgery.

Exclusion criteria included:
Sublobar resection (e.g., wedge resection or segmentectomy).

Incomplete clinicopathological or follow-up data.

Non-R0 resections or inadequate lymph node sampling.

Receipt of fewer than four cycles of ACT.

History of other malignancies.

A total of 276 eligible patients with complete histopathological and follow-up data were included in the final analysis (Figure 1).

Histopathologic evaluation
All resected specimens were processed according to standard clinical protocols and independently reviewed by two experienced pulmonary pathologists, as noted above. All pathologists were blinded to clinical outcomes and follow-up data. Histopathological evaluation followed the 2020 IASLC classification system for invasive adenocarcinoma. The presence and proportion of each histologic component were recorded from pathology reports using semi-quantitative estimates in 5% increments, and the predominant histologic subtype was defined as the component occupying the largest proportion.
Clinical and pathological variables collected included age at surgery, sex, smoking history, tumor location, surgical approach, histologic subtype and grade, tumor size, VPI status, chemotherapy regimen, survival status, and follow-up time.

Statistical analysis
Patients were divided into two groups based on whether they received ACT. The primary endpoints were DFS and OS. DFS was defined as the time from surgery to tumor recurrence, death from lung cancer, or the last follow-up. OS was defined as the time from surgery to death from any cause or the last follow-up, whichever occurred first.
Continuous variables were compared using Student’s t-test, while categorical variables were analyzed using the chi-squared or Fisher’s exact test, as appropriate. Propensity score matching (PSM) was performed to minimize selection bias between the ACT and non-ACT groups. Matching was performed at a 1:1 ratio using the nearest-neighbor method without replacement, based on age, sex, smoking status, tumor size, histologic subtype, and high-risk features (VPI and LVI). Covariate balance before and after matching was evaluated using both P values and standardized mean differences (SMDs), with SMD <0.10 considered indicative of good balance.
Survival curves for DFS and OS were generated using the Kaplan-Meier method and compared via the log-rank test. Multivariate Cox proportional hazards models were then fitted within the post-PSM matched cohort to estimate the independent association between ACT and survival outcomes. Robust sandwich (cluster-robust) variance estimators were used to account for matched-pair dependence (cluster = subclass). The proportional hazards assumption was verified using Schoenfeld residuals. Subgroup analyses were conducted according to histologic grade, tumor size, and VPI status. A two-sided P value of <0.05 was considered statistically significant. All statistical analyses were performed using R software (version 4.2.1, R Foundation for Statistical Computing, Vienna, Austria).

Results

Patient characteristics
A total of 276 patients with stage IB lung adenocarcinoma were included in the study, of whom 121 (43.8%) received ACT and 155 (56.2%) did not. The median follow-up duration was 66.3 months (range, 7–108 months). The clinical data of these patients are summarized in Table 1.
Before matching, significant differences were observed between the ACT and non-ACT groups in several baseline characteristics, including age and VPI. After PSM, 106 matched pairs were generated, and baseline characteristics were generally well balanced between the two groups (Table 2). In addition to P value comparisons, SMDs were calculated to further evaluate matching adequacy. Most covariates had SMDs <0.10 after matching, indicating excellent balance, while age (SMD =0.251), tumor diameter (SMD =0.181), and VPI (SMD =0.150) showed modest residual imbalance (Table S1, Figure S1). These variables were therefore retained as covariates in the multivariable Cox regression models to control for potential residual confounding.
Disease recurrence occurred in 74 patients (26.8%), including 26 (35.1%) in the ACT group and 48 (64.9%) in the non-ACT group, confirming that the two groups were sufficiently comparable for subsequent survival analyses

Survival outcomes
ACT significantly improved both DFS and OS. Multivariate Cox analysis confirmed ACT as an independent protective factor for DFS [hazard ratio (HR) =0.471, 95% confidence interval (CI): 0.319–0.694; P=0.001] and OS (HR =0.519, 95% CI: 0.287–0.939; P=0.03) in the matched cohort (cluster-adjusted models). Kaplan-Meier curves demonstrated better DFS (P=0.002) and OS (P=0.02) in the ACT group (Figure 2).

Prognostic factors
Independent risk factors for poor DFS included high histologic grade (HR =3.007, P=0.005), predominantly solid (HR =6.140, P=0.001), micropapillary (HR =8.865, P=0.001), or complex glandular patterns (HR =6.174, P=0.001), and LVI (HR =1.753, P=0.048). For OS, older age (HR =1.087, P=0.001), larger tumor size (HR =2.167, P=0.008), VPI (HR =2.299, P=0.02), and the same histologic patterns were significant adverse prognostic factors. The results are shown in Table 3.

Subgroup analysis
The benefit of ACT was histologic grade-dependent. In patients with poorly differentiated (G3) tumors, ACT reduced recurrence risk by 66.5% (HR =0.335, P=0.001), whereas no benefit was observed in moderately differentiated (G2) tumors (HR =1.041, P=0.91). A significant treatment-grade interaction (HR =2.799, P=0.03) confirmed histologic grade as an effect modifier.
Further subgroup analysis by VPI status showed that ACT improved DFS in both VPI-positive (HR =0.535, P=0.04) and VPI-negative (HR =0.494, P=0.052) patients, although statistical significance was only reached in the VPI-positive group. The interaction between ACT and VPI was not significant (HR =1.029, P=0.95), suggesting consistent benefit regardless of VPI status (Figure 3).

Discussion
The role of ACT in stage IB NSCLC remains controversial (20,21). Previous studies have rarely focused specifically on high-risk subgroups to evaluate the true benefit of ACT (18,22). In this study, we evaluated patients with stage IB NSCLC who underwent complete resection and presented with at least one high-risk feature, such as VPI, LVI, or poor differentiation. Our findings demonstrate that ACT significantly improves survival in high-risk stage IB NSCLC (HR =0.471, 95% CI: 0.319–0.694; P=0.001), helping to reconcile inconsistencies among current treatment guidelines. Importantly, the benefit of ACT was largely confined to G3 tumors, in which it reduced the risk of recurrence by 66.5% (HR =0.335, 95% CI: 0.197–0.567; P=0.001). In contrast, G2 tumors derived no significant benefit (HR =1.041, 95% CI: 0.518–2.090; P=0.91), as confirmed by a significant treatment-grade interaction (HR =2.799, 95% CI: 1.137–6.890; P=0.03). By contrast, VPI status did not significantly influence the efficacy of ACT (interaction HR =1.029, 95% CI: 0.395–2.683; P=0.95), supporting the use of ACT in high-risk patients irrespective of VPI status. Taken together, these findings help reconcile guideline discrepancies by showing that ACT is effective when selectively applied to high-risk subgroups.
Histologic components have long been recognized as important prognostic indicators influencing the efficacy of chemotherapy in lung adenocarcinoma (4,17-19,23,24). Qian et al. reported that ACT conferred survival benefits in patients with solid/micropapillary-predominant (SMPP) stage IB adenocarcinoma, whereas no such benefit was observed in those with only minor solid/micropapillary components (SMPM; >5% but not predominant) (3). Similarly, Cao et al. found that ACT improved DFS and OS in patients with solid-predominant stage IB tumors, but provided no survival advantage in those with non-solid-predominant patterns (19). However, few studies have applied the 2020 IASLC histologic grading system, resulting in continued controversy over the prognostic and therapeutic relevance of low tumor differentiation as a high-risk feature.
In our study, histologic grade was a significant predictor of DFS in both univariate and multivariate analyses. However, for OS, histologic grade was associated with prognosis only in univariate analysis, and this association did not remain significant after adjusting for confounding variables in the multivariate model. Thus, histologic grade appears to be a strong predictor of recurrence in stage IB NSCLC, supporting its role in guiding ACT decisions to reduce early treatment failure. However, its impact on long-term survival warrants further validation through extended follow-up and large-scale pooled analyses.
Subgroup analysis confirmed that ACT improves survival in high-risk stage IB NSCLC, but importantly, this benefit was concentrated in patients with G3 tumors. A significant treatment-grade interaction (HR =2.799, 95% CI: 1.137–6.890; P=0.03) further supports histologic grade as an effect modifier of ACT efficacy. While ACT reduced the risk of recurrence by 66.5% in G3 patients (HR =0.335, 95% CI: 0.197–0.567; P=0.001), it conferred no measurable benefit in G2 tumors (HR =1.041, 95% CI: 0.518–2.090; P=0.91), despite both subgroups sharing the same anatomical high-risk features.
These findings suggest a need to refine current definitions of high-risk status in stage IB NSCLC. The NCCN guidelines broadly recommend ACT for poorly differentiated stage IB NSCLC, whereas the CSCO guidelines take a more conservative stance, discouraging routine use due to insufficient supporting evidence. Our findings help reconcile this discrepancy: the NCCN recommendation appears appropriate for G3 tumors, where ACT confers significant benefit, while the CSCO perspective aligns with G2 cases, where the benefit is negligible and overtreatment is a concern.
For clinical practice, we propose a stratified approach: G3 + High-risk features: strong ACT recommendation; G2 + High-risk features: individualized decision-making weighing ACT toxicity against marginal benefit. This strategy prevents G3 undertreatment while sparing G2 patients unnecessary toxicity—addressing a core dilemma in IB NSCLC management.
In addition to the widely recognized solid and micropapillary components, our study also incorporated complex glandular patterns (CGPs) as predominant histologic subtypes, following the 2020 IASLC grading system. Our results demonstrated that CGPs, along with solid and micropapillary patterns, served as independent adverse prognostic factors for both DFS and OS (25,26).
Although VPI is widely recognized as a high-risk feature in early-stage NSCLC (27), our study found that it did not significantly modify the efficacy of ACT. ACT improved DFS in both VPI-positive (HR =0.535, P=0.04) and VPI-negative (HR =0.494, P=0.052) subgroups, with no significant interaction observed (P=0.95). These findings suggest that the therapeutic benefit of ACT in high-risk stage IB adenocarcinoma is not contingent on VPI status.
This contrasts with the LACE meta-analysis, which reported enhanced ACT benefit in patients with VPI (28). Several factors may explain this discrepancy. First, the LACE analysis included a more heterogeneous population, with only 45% of patients having stage IB disease and 40% exhibiting squamous histology, which may respond differently to ACT. Second, evolving treatment paradigms—such as the increasing use of targeted therapies in EGFR-mutant tumors—may attenuate the relative contribution of ACT, particularly in molecularly selected patients (29). Finally, differences in tumor biology and limited sample sizes in VPI-negative cohorts may also contribute to inconsistent results.
Taken together, our findings suggest that while VPI remains a relevant prognostic factor, it may not serve as a reliable predictor of ACT benefit in modern stage IB adenocarcinoma. Clinical decision-making should therefore incorporate histologic grade and molecular characteristics, rather than relying on VPI alone.
Although targeted therapies have become integral to the precision treatment of lung cancer (29-31), our study reaffirms the continued relevance of traditional chemotherapy in stage IB patients—particularly those with high-risk features who are ineligible for, or resistant to, molecular targeted agents (32). Recent molecular evidence further indicates that the efficacy of ACT may vary according to oncogenic driver status. Aoki et al. demonstrated that postoperative tegafur-uracil (UFT) significantly improved survival in EGFR wild-type stage IB adenocarcinoma (33). Park et al. highlighted the uncertain benefit of chemotherapy prior to adjuvant osimertinib, underscoring the complexity of treatment sequencing in this subgroup (34).
Collectively, these observations indicate that pathological aggressiveness and molecular genotype exert independent yet complementary effects on postoperative outcomes. Integrating both dimensions in future prospective analyses could refine postoperative risk stratification and enable more biologically informed adjuvant treatment selection for early-stage lung adenocarcinoma.
There are several inherent limitations in this study, primarily due to its single-center, retrospective design. First, selection bias and treatment allocation bias are inevitable in retrospective studies, even with the use of PSM. Specifically, the exclusion of patients who received fewer than four cycles of ACT may have further introduced selection and immortal-time biases, potentially leading to an overestimation of ACT benefit, as the analyzed ACT group was enriched with patients who were robust enough to complete the full treatment course. Second, the absence of a low-risk control group limits our ability to determine whether the current high-risk criteria effectively discriminate patients who truly benefit from ACT. Validation in external cohorts—including stage IB patients without high-risk features—is therefore needed. Third, relatively small subgroup sample sizes may have limited the statistical power to detect modest treatment effects, particularly in heterogeneous populations.

Conclusions
This study demonstrates that ACT significantly improves DFS and OS in stage IB NSCLC patients with high-risk features, particularly those with G3 tumors. Histologic grade was the key modifier of ACT efficacy, while VPI status did not significantly alter treatment benefit. Pathological patterns such as solid, micropapillary, and complex glandular components, as well as LVI, were independent predictors of poor prognosis. These findings support a risk-stratified approach to ACT, favoring its use in G3 patients while advocating individualized decision-making in G2 cases. Further prospective, multicenter studies incorporating molecular profiling are warranted to validate this strategy and refine treatment selection.

Supplementary
The article’s supplementary files as