This two-centric, retrospective cohort study included patients who met the following criteria: pathologically unresectable stage III NSCLC; ALK-positive status was identified by one of Ventana D5F3 immunohistochemistry (IHC), fluorescent in situ hybridization (FISH), reverse transcription-polymerase chain reaction (RT-PCR) or next-generation sequencing (NGS); and adequate follow-up data. The follow-up period ranged from the date of diagnosis to either the last follow-up or death. Participants were followed every 6 months through clinic visits or telephone interviews.

Patients were categorized into two treatment groups: the definitive radiation therapy (RT) group and the systemic therapy group. Definitive RT was defined as volumetric modulated arc therapy (VMAT) or intensity-modulated radiation therapy (IMRT) with a prescribed dose of 60–70 Gy, delivered sequentially or concurrently with at least two cycles of platinum-based doublet chemotherapy. The systemic therapy group comprised patients who were unsuitable for definitive RT or those who refused it, opting for drug therapy as their initial treatment.

NGS was used to detect the ALK rearrangements. EML4-ALK v3 was defined as the fusion of exon 6a/b in EML4 to exon 20 in ALK, while non-EML4-ALK v3 variants included other EML4–ALK fusions (e.g., EML4–ALK v1, v2) and additional ALK fusion partners such as TMEM178A-ALK and KCNK2-ALK. Co-mutation was defined as ALK-positive status combined with other mutation, such as EGFR and BRCA2. Gene ontology (GO) enrichment analysis was performed on data from 25 patients who underwent NGS.

GO enrichment analysis was performed to investigate the cellular functions of smoking-related genes. Smoking-related genes were defined as differentially expressed genes between smokers and never-smokers in the GSE31852 cohort, with an absolute log2 fold change (|log2FC|) > 2 and P < 0.05, yielding 89 genes. The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed. Analysis was conducted using the online Database for Annotation, Visualization, and Integrated Discovery (DAVID 6.8, https://david.ncifcrf.gov/) to explore the potential biological roles of smoking-related genes. Gene set variation analysis (GSVA) was performed using R 3.5.1 to estimate variation in gene set enrichment according to the expression data, using the GSE31852 dataset (n = 34) from the Gene Expression Omnibus (GEO) database. The GSVA package was freely available at http://www.bioconductor.org.

Progression-free survival (PFS), locoregional-free survival (LRFS), distant metastasis-free survival (DMFS), and overall survival (OS) were defined as the time from diagnosis to disease progression or death, locoregional recurrence or death, distant recurrence or death, and death, respectively. Kaplan–Meier curves were used to estimate survival outcomes. Baseline characteristics between the groups were compared using the Chi-square test or Fisher’s exact test for categorical variables, as appropriate. Landmark analysis was employed to reduce the impact of long-term events on outcomes. Univariate and multivariate regression analyses were carried out using a Cox proportional hazards model, and covariates with P<0.2 in the univariate analysis were included in the multivariate analysis. All statistical analyses were performed using R software (version 4.3.2), with two-sided P < 0.05 was considered statistically significant.

A total of 48 patients with stage III unresectable ALK-positive NSCLC met the eligibility criteria and were included in the analysis (Supplementary Figure S1). The baseline characteristics of patients are summarized in Table 1. The median age was 55.2 (33–80) years; almost half of the patients were men (50.0%) and current or former smokers (45.8%); 62.5% had stage IIIB disease; most patients (93.8%) had adenocarcinoma; all patients had an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0-1. 16 (33.3%) patients received definitive CRT as the initial treatment; 14 (29.2%) patients had EML4-ALK v3 status. Regarding treatment-related toxicities, pneumonitis was observed in 6/16 patients in the RT group, including one case of grade 3 case.

The median follow-up time for the entire cohort was 49.02 (interquartile range [IQR], 35.84 -62.03) months. The 1-, 3-, 5-year OS, and median OS for all patients were 91.6% (95% confidence interval [CI]: 84.0% - 99.8%), 70.7% (95% CI: 58.4% - 85.6%), 67.0% (53.8% - 83.4%) and not reached, respectively; the 1-, 3-, 5-year PFS, and median PFS was 58.3% (95% CI: 45.9% - 74.1%), 24.1% (95% CI: 14.0% - 41.4%), 15.1% (95% CI: 6.7% - 35.8%) and 14.7 months (95% CI: 10.6-24.9), respectively (Supplementary Figure S2). Compared with systemic therapy, the Kaplan-Meier curves showed that CRT improved the mPFS (21.1 vs. 9.6 months, P = 0.065; Figure 1B) and mLRFS (NA vs. 14.4 months, P = 0.012; Figure 1C). In contrast, no significant difference was observed in OS or DMFS between the CRT and systemic therapy. Subsequently, we examined the effect of smoking status on OS, PFS, LRFS, and DMFS (Figure 2). The 5-year OS rate was significantly higher in the never-smokers group at 78.9% (95% CI: 63.9% - 97.3%) compared to 51.9% (95% CI: 32.6% - 82.8%) in the smokers group (P = 0.038; Figure 2A). Similar trends were observed for DMFS, with a 5-year DMFS of 45.0% (95% CI: 26.6% - 76.4%) versus 7.5% (95% CI: 1.2% - 45.6%) in the never-smokers and smokers groups, respectively (P = 0.004; Figure 2D).

As summarized in Supplementary Table S2, 77.1% of patients had relapsed. 15 (31.3%) patients had distant recurrence only and 13 (27.1%) patients had locoregional and distant recurrence simultaneously. Brain (14/28) was the most common metastasis site. Compared with systemic therapy, CRT significantly decreased the locoregional recurrence rate (18.8% vs. 59.4%, P = 0.018, Supplementary Table S2).

The univariate Cox regression analysis revealed that smoking status (hazard ratio [HR] = 3.01, 95% CI: 1.01 - 9.02, P = 0.049) and stage (IIIB vs. IIIA, HR = 0.18, 95% CI: 0.06 - 0.53, P = 0.002) were significantly associated with OS (Table 2). In the univariate Cox regression analysis, the variable stage IIIC produced an infinite HR due to the absence of observed deaths in this category, which this variable was excluded from the multivariate Cox regression analysis to ensure model stability and reliability. In the multivariate Cox regression analysis, after adjusting for confounders, smoking status (HR = 3.92, 95% CI: 1.21 - 12.73, P = 0.023) remained a significant predictor of OS (Table 2), indicating smoking status was the independent prognostic factor for OS in stage III ALK-positive NSCLC. As for PFS (Table 3), univariate COX analysis suggested that patients who received systemic therapy tended to have a poorer PFS (HR = 1.92, 95%CI 0.95 - 3.89 P = 0.069). However, after adjusting for potential confounders (Covariates with P<0.2 in the univariate analysis) in the multivariate analysis, treatment was found to be significantly associated with PFS (aHR: 3.04, 95% CI: 1.32 - 7.01, p = 0.009), which indicates CRT exerts a significant effect on PFS when accounting for other covariates.

To further investigate the differences between smokers and never-smokers, we performed GO analyses. As illustrated in Supplementary Figure S3, the results revealed distinct revealed distinct enrichment patterns. For biological processes (BP), never-smokers exhibited enrichment in peptidyl-tyrosine modification, positive regulation of kinase activity, and peptidyl-tyrosine phosphorylation, whereas smokers showed enrichment in gland development, epithelial cell proliferation, and regulation of the mitotic cell cycle. In terms of cellular components (CC), never-smokers were enriched in transcription repressor complex, transcription regulator complex, and PML body, whereas smokers were enriched in chromosome, telomeric region, DNA repair complex, and PML body. For molecular functions (MF), both smokers and never-smokers demonstrated enrichment in transmembrane protein kinase activity, protein serine/threonine/tyrosine kinase activity, and transmembrane receptor protein tyrosine kinase activity. Additionally, KEGG pathway analyses indicated that smokers were particularly enriched in pathways related to cellular senescence.

The GSE31852 cohort was used for subsequent analysis to explore the differences between smokers and never-smokers. Initially, 89 differentially expressed genes were identified. Figure 3B shows a heatmap presenting the detailed expression of smoking-related genes. GO analysis indicated that these genes were enriched in lymphocyte differentiation, T cell selection, and positive T cell selection in the BP category; external side of plasma membrane, spindle, and alpha-beta T cell receptor complex in the CC category; and heparin binding and signaling receptor complex adaptor activity in the MF category. KEGG enrichment analysis revealed that these smoking-related genes might be associated with measles, hematopoietic cell lineage, and primary immunodeficiency (Figure 3C). To gain a deeper understanding of the mechanisms underlying these differentially expressed genes, we utilized GSEA and found that the genes were related to immunoregulatory interactions between a lymphoid and a non-lymphoid cell, as well as cell cycle checkpoints.