S-1 is an oral fluoropyrimidine formulation combining tegafur, a prodrug of 5-fluorouracil (5-FU), with gimeracil, which inhibits 5-FU degradation and enhances antitumor activity, and oteracil potassium, which reduces the gastrointestinal toxicity of 5-FU by inhibiting its phosphorylation in the gastrointestinal tract (1). It was approved in Japan in 2004 for the treatment of non-small-cell lung cancer (NSCLC). The antitumor activity of S-1 monotherapy or combination regimens as first-, second-, and later-line treatments was demonstrated in patients with advanced NSCLC (2-6). Based on those findings, S-1 has been incorporated into Japanese clinical practice guidelines as a treatment option for advanced NSCLC (7).

However, in the era of immune checkpoint inhibitors and molecularly targeted therapies, the clinical role of S-1 has become less clearly defined (8). Nevertheless, it continues to be used in clinical practice, particularly for patients with a manageable toxicity profile who require outpatient therapy (7). Although S-1 can achieve durable disease control in a subset of patients with advanced NSCLC, clinical factors associated with prolonged treatment remain poorly understood. Furthermore, real-world data on treatment duration and subsequent outcomes remain limited, particularly in heavily pretreated populations. Therefore, we conducted a retrospective study to evaluate the clinical characteristics and outcomes of patients treated with S-1 monotherapy for advanced NSCLC, with a particular focus on factors associated with prolonged treatment.

Patients. We retrospectively reviewed the data of patients with advanced NSCLC who started S-1 monotherapy at Kanazawa Medical University Hospital from January 2015 to December 2025. Eligible patients had stage IIIB, stage IV, or postoperatively recurrent NSCLC. They were included regardless of treatment line, provided that S-1 was administered as monotherapy. The cut-off date for collection of clinical data for this analysis was mid-May 2026.

Treatment and definition of treatment duration. S-1 was administered according to a 2-week-on/1-week-off or 4-week-on/2-week-off schedule. Treatment schedules were modified during therapy based on tolerance and the patient’s clinical condition.

The treatment duration was defined as the time from the first to the last administration of S-1, regardless of schedule modifications. Patients were categorized into a long-term treatment group (≥90 days) or a short-term treatment group (<90 days).

Response assessment. The treating physicians assessed the tumor response via computed tomography according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 (9). The best overall response during S-1 treatment was recorded as a complete response, partial response, stable disease, progressive disease, or not evaluable.

Statistical analysis. Statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Categorical variables were compared using the chi-square test or Fisher’s exact test, as appropriate. Continuous variables were compared using Student’s t-test. Overall survival (OS) was defined as the time from initiation of S-1 monotherapy for advanced NSCLC to death from any cause and was estimated using the Kaplan–Meier method. Differences between groups were assessed using the log-rank test. All tests were two-sided, and a value of p<0.05 was considered statistically significant.

As this was a retrospective exploratory study in which we included all consecutive eligible patients during the study period, no formal sample size calculation was performed. Baseline characteristics were summarized descriptively, and no formal hypothesis testing was performed for these variables. Statistical comparisons were performed for the disease control rate, treatment duration, and subsequent therapy, as these outcomes were predefined, clinically relevant endpoints.

Ethics statement. This study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Kanazawa Medical University Hospital (approval no. I812; approved on May 10^th, 2023). Because of the retrospective nature of the study, information was disclosed using an opt-out approach instead of obtaining written informed consent.

The study included 32 patients with advanced NSCLC. The median duration of S-1 treatment was 103.4 days (range=12-342 days). Thirteen patients (40.6%) were classified into the long-term treatment group, and 19 (59.4%) into the short-term treatment group. At data cut-off, three patients were still receiving treatment. Tumor response evaluation revealed a partial response in 4 patients, stable disease in 11, progressive disease in 15, and an unevaluable response in 2, yielding an objective response rate of 12.5% and a disease control rate of 46.9%.

Baseline clinical, pathological, and treatment-related characteristics are summarized in Table I. No significant differences were observed between the two groups in terms of age, sex, smoking status, histology, programmed death-ligand 1 expression, epidermal growth factor receptor (EGFR) mutation status, metastatic sites, treatment line, prior immune checkpoint inhibitor exposure, the neutrophil-to-lymphocyte ratio, or tumor size. EGFR mutations were detected in six patients, including exon 19 deletion (n=4), the L858R variant (n=1), and exon 20 insertion (n=1); their proportions were well balanced between the two groups.

Regarding baseline radiological patterns, only one patient in the long-term group exhibited honeycombing consistent with a usual interstitial pneumonia (UIP) pattern, while no definite honeycombing was observed in the other patients. The prevalence of pre-existing interstitial lung disease (ILD) did not differ significantly between groups.

Treatment outcomes and adverse events are summarized in Table II. Patients in the long-term treatment group received S-1 for a median of 189 days (range=90-342 days), whereas those in the short-term treatment group received treatment for a median of 45 days (range=12-89 days). The disease control rate was significantly higher in the long-term treatment group than in the short-term treatment group (84.6% vs. 21.1%, p=0.001). The disease control rate was significantly higher for the long-term treatment group (p=0.001). Treatment-related adverse events were generally mild. Grade 1 toxicities included lacrimation in two patients. Grade 2 toxicities were observed in five patients (appetite loss in three, diarrhea in one, and thrombocytopenia in one). No grade ≥3 toxicities or treatment-related deaths occurred. Kaplan–Meier analysis revealed that the long-term treatment group had a significantly longer OS than the short-term treatment group (median=4,348 vs. 1,557 days; p=0.003; Figure 1).

In this retrospective study of S-1 monotherapy for patients with advanced NSCLC, prolonged treatment (≥90 days) was associated with significantly longer OS than shorter treatment. Although baseline characteristics were well balanced between the two groups, the disease control rate was significantly higher for the long-term treatment group, suggesting a relationship between treatment continuation and favorable clinical outcomes.

Importantly, although the difference in performance status between the groups did not reach statistical significance, poor performance status (Eastern Cooperative Oncology Group performance status of 2-3) was more common in the short-term treatment group, whereas none of the patients in the long-term treatment group had such a poor performance status. This finding suggests that baseline functional status may be an important determinant of treatment continuation with S-1. Patients able to tolerate prolonged S-1 therapy may represent a clinically favorable subgroup characterized by a preserved performance status and more indolent disease biology, which may partially explain the observed survival differences.

Furthermore, although the difference was not statistically significant, a greater proportion of patients in the long-term treatment group received subsequent systemic therapy than those in the short-term treatment group. This finding may indicate that the functional status was better maintained among patients who were able to continue S-1 treatment for a longer period. By contrast, patients in the short-term treatment group were more likely to discontinue S-1 therapy because of early disease progression or clinical deterioration, potentially limiting opportunities for further treatment.

The objective response rate observed in this study was modest, consistent with those in previous reports of S-1 monotherapy in patients previously treated for NSCLC (10, 11). These findings suggest that disease stabilization, rather than tumor shrinkage, may be the principal clinical benefit of S-1 therapy.

In addition, a higher proportion of patients with thyroid transcription factor 1 (TTF1)-positive tumors was observed in the long-term treatment group, although the difference was not statistically significant. Previous studies have suggested that S-1 may demonstrate greater clinical activity in non-squamous NSCLC, particularly adenocarcinoma, than in squamous cell carcinoma (12, 13). Previous study has also indicated that TTF1 expression status may be associated with distinct clinical characteristics and treatment outcomes in non-squamous NSCLC (14). These findings suggest that tumor biological features, including histological subtype and TTF1 expression status, may contribute to heterogeneity in response to systemic therapy, including S-1. Therefore, the higher proportion of TTF1-positive tumors in the long-term treatment group may partly reflect differences in underlying tumor biology rather than a direct predictive effect of TTF1 expression itself.

However, given the limited sample size, retrospective study design, and lack of important pharmacological biomarkers, such as thymidylate synthase expression (15), the clinical significance of these findings remains uncertain. Therefore, these observations should be considered hypothesis-generating, and further studies are warranted to clarify the potential relationship between tumor biology and sensitivity to S-1 monotherapy.

Regarding safety, S-1 was generally well tolerated in this real-world cohort. The grades of treatment-related adverse events were 1-2, and no grade 3 or higher toxicities or treatment-related deaths were observed. Pre-existing ILD was present in a subset of patients, but it was not associated with the treatment duration. Importantly, no cases of drug-induced ILD or acute exacerbation of pre-existing ILD occurred during S-1 treatment. Although rare cases of S-1-associated ILD have been reported (16), previous studies have suggested that S-1-based regimens may be feasible for patients with NSCLC complicated by ILD (17, 18). These findings support the feasibility of S-1 monotherapy for carefully selected patients, including those with coexisting pulmonary disease.

Overall, these findings suggest that the clinical benefit of S-1 monotherapy for advanced NSCLC may be associated more closely with disease stabilization and the ability to maintain treatment exposure than with tumor shrinkage alone. Patients who can tolerate long-term treatment may represent a subgroup characterized by preserved functional status and more indolent disease biology.

In this context, pretreatment tumor growth kinetics may represent another potential factor reflecting underlying tumor characteristics. Our previous study demonstrated an association between pretreatment tumor growth rate and clinical outcomes in advanced NSCLC, suggesting that tumor growth behavior may contribute to differences in treatment response and survival outcomes (19).

The optimal duration of anticancer therapy remains an important clinical question across different cancer types. Previous study has investigated the relationship between treatment duration and clinical outcomes in other malignancies (20), although the appropriate duration may vary depending on the disease setting and treatment intent. These findings highlight the importance of balancing treatment continuation with tolerability and disease control, particularly in the setting of long-term systemic therapy.

Therefore, careful patient selection based on both performance status and underlying tumor characteristics may be important when S-1 is considered as a later-line treatment option.

Study limitations. Firstly, the retrospective design and small sample size limit the statistical power and generalizability of the results. Secondly, selection bias was unavoidable because patients who were able to continue treatment for longer periods may have inherently possessed more favorable clinical and biological characteristics. Thirdly, heterogeneity in prior systemic therapies, including immune checkpoint inhibitors, may have influenced the observed outcomes. Finally, classification based on treatment duration introduces the possibility of immortal time bias; therefore, the observed survival differences should be interpreted as an association rather than as evidence of causality. Accordingly, the association between prolonged S-1 treatment and improved survival observed in this study should be interpreted as an observational finding potentially influenced by selection bias.

Prolonged S-1 treatment of patients with advanced NSCLC was associated with improved OS and disease control without new safety concerns. These findings suggest that S-1 monotherapy may provide a clinically meaningful benefit in selected patient subgroups in real-world practice.

The Authors declare that they have no conflicts of interest related to this study.

Conceptualization: YT. Resources: YT, RA, SN, TT, YI, IS, KY and MI. Investigation: RA, SN, TT, YI, IS, KY and MI. Methodology: YT and MI. Writing - original draft: YT, with support from MI. All Authors had full access to the study data and take responsibility for the integrity of the data and the accuracy of the data analysis. All the Authors read and approved the submission of this manuscript.

During the preparation of this manuscript, a large language model (ChatGPT, OpenAI) was used solely for language editing and stylistic improvements in select paragraphs. No sections involving the generation, analysis, or interpretation of research data were produced by generative AI. All scientific content was created and verified by the authors. Furthermore, no figures or visual data were generated or modified using generative AI or machine learning–based image enhancement tools.