This study retrospectively analysed data from a cohort of patients diagnosed with non-small cell lung cancer (NSCLC) with lymph node metastasis at Xi’an International Medical Center Hospital between April 2019 and October 2022. The study protocol was approved by the Ethics Committee of Xi’an International Medical Center Hospital (Approval No. 2024034) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients or their legal guardians.

Patient selection followed a structured process. Initially, 155 patients who met the preliminary diagnostic criteria were identified from hospital records, including 70 patients who received ICI and 85 patients treated with chemotherapy. Propensity score matching (PSM) was subsequently performed in a 1:1 ratio, yielding a final analysis cohort of 112 patients, with 56 assigned to each group (ICI and chemotherapy).

Patients were eligible for the study if they met the following criteria: age $\ge$18 years; histologically confirmed NSCLC with lymph node metastasis [3]; disease progression or intolerance after at least one platinum-based chemotherapy regimen; unresectable disease scheduled for immunotherapy or chemotherapy; actual survival of more than 3 months; Eastern Cooperative Oncology Group (ECOG) performance status of 0–1 [9]; and at least one measurable lesion as defined by the Response Evaluation Criteria in Solid Tumours (RECIST), version 1.1 [10].

Exclusion criteria included: prior surgical treatment for NSCLC; concurrent malignancies or severe interstitial lung disease; presence of autoimmune diseases or requirement for corticosteroid or immunosuppressive therapy; and any other contraindications to immunotherapy or chemotherapy.

The final matched cohort was divided into two treatment groups. The ICI group (n = 56) received pembrolizumab monotherapy (JS20180031, MSD Ireland, Berlin, Germany) at a fixed dose of 200 mg administered intravenously every 21 days. The chemotherapy group (n = 56) received docetaxel monotherapy (H20020543, Hengrui Co., Ltd., Nanjing, China) at 75 mg/m², administered every 21 days.

Clinical variables, including age, gender, histological subtype, ECOG performance status, PD-L1 expression level, smoking history, tumour-node-metastasis (TNM) staging (N classification), epidermal growth factor receptor‌ (EGFR) and anaplastic lymphoma kinase (ALK) mutation status, and the number of prior systemic therapy lines, were retrospectively retrieved from the electronic medical records of the hospital.

The primary efficacy endpoint was the objective response rate (ORR). Secondary endpoints included the disease control rate (DCR) and survival outcomes.

Efficacy assessment: tumour response was radiologically evaluated every two treatment cycles (approximately every 6 weeks) according to RECIST version 1.1 [10]. Tumour responses were classified as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). ORR was defined as the proportion of patients achieving CR or PR:

$$\text{ORR}(\%)=[(\text{CR}+\text{PR})/\text{total number}]\times 100.$$

DCR was defined as the proportion of patients achieving CR, PR, or SD:

$$\text{DCR}(\%)=[(\text{CR}+\text{PR}+\text{SD})/\text{total number}]\times 100.$$

Survival follow-up: from treatment initiation, patient survival status was evaluated every three months through outpatient visits or telephone interviews. Follow-up continued until death or the last recorded contact.

Safety assessment: adverse events (AEs) were monitored from treatment initiation until 30 days after the final dose. Collected safety data included common AEs (e.g., anaemia, leukopenia, thrombocytopenia, nausea, vomiting, and elevated transaminases) as well as immune-related adverse events (irAEs). All AEs were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) [11].

Statistical analyses were performed using SPSS software (version 26.0, IBM Corp., New York, NY, USA), while graphical presentations were generated with GraphPad Prism (version 6.0, GraphPad Prism Software Inc., San Diego, CA, USA). A two-sided p-value 0.05 was considered statistically significant.

To minimise selection bias and balance baseline characteristics, 1:1 nearest-neighbour propensity score matching (PSM) was applied. Matching variables included age, gender, histological subtype, ECOG performance status, PD-L1 expression level, smoking history, number of prior systemic therapies, N stage status, EGFR mutation status, and ALK rearrangement status. The calliper width was set at 0.05. In addition to p-values, standardized mean differences (SMDs) were calculated to assess the balance of baseline covariates between groups. An SMD 0.2 was considered to indicate negligible imbalance.

Categorical variables are presented as counts (n) and proportions (%). Group comparisons were selected based on the minimum expected frequency (T) in contingency tables: the chi-square test was applied if T $\ge$5; the chi-square corrected test if 1 $\le$ T 5; and Fisher’s exact test if T 1. The chi-square test was employed to compare ORR and DCR between groups. Survival outcomes were estimated using the Kaplan-Meier method and compared with the log-rank test. Independent prognostic factors for overall survival (OS) were identified using univariate and multivariate Cox proportional hazards regression analyses. Variables of clinical significance, including treatment group, age, ECOG performance status, and PD-L1 expression, were entered into the multivariate model to assess their independent effects on survival, irrespective of their significance in univariate testing. For subgroup analyses of objective response rate (ORR), treatment efficacy was assessed by calculating odds ratios (ORs) with 95% confidence intervals (CIs) derived from logistic regression models for each predefined subgroup.

A total of 155 patients with unresectable NSCLC and lymph node metastasis were initially identified, including 85 patients in the chemotherapy group and 70 in the ICI group. Following 1:1 PSM, a final matched cohort of 112 patients was established, with 56 participants in each treatment group (Fig. 1). Detailed demographic and clinical characteristics of patients before and after PSM are presented in Table 1. Before matching, ECOG performance status differed significantly between the two groups (p = 0.017). However, other baseline variables, including age, gender, histological subtype, PD-L1 expression, smoking history, number of prior therapies, TNM stage, and driver gene status (EGFR/ALK), were comparable between groups (all p $>$ 0.05). After PSM, the imbalance in baseline characteristics was substantially reduced, as reflected by decreased SMD values across most covariates. No statistically significant differences were observed between the ICI and chemotherapy groups (all p $>$ 0.05).

Fig. 1.

Patient selection flowchart. NSCLC, non-small cell lung cancer; ICI, immune checkpoint inhibitor.

Treatment outcomes, including ORR and DCR, are summarised in Table 2. The ORR was significantly higher in the ICI group compared with the chemotherapy group (p = 0.012). Conversely, the DCR did not differ significantly between the two groups (p = 0.057).

As shown in Fig. 2, subgroup analysis demonstrated that ICI therapy was associated with higher ORR compared with chemotherapy across several patient subgroups. Notably, patients with PD-L1 expression $\ge$50%, ECOG performance status of 1, N2-stage disease, and age $\ge$65 years exhibited significantly greater benefit from ICI treatment. Additionally, a trend favouring ICI over chemotherapy was observed among patients with PD-L1 expression of 1–49%, age 65 years, and ECOG performance status of 0, although these differences did not reach statistical significance.

Fig. 2.

Forest plot of objective response rate (ORR) by subgroup (ICI vs. Chemotherapy). TNM, tumour-node-metastasis; PD-L1, Programmed Cell Death Ligand 1.

Patients were followed for a median duration of 24 months. Overall survival (OS) was estimated using the Kaplan-Meier method, and survival curves for the ICI and chemotherapy groups are presented in Fig. 3. During the 24-month follow-up period, patients in the ICI group consistently demonstrated a higher survival probability compared with those in the chemotherapy group. The 1-year survival rates were 76.8% for the ICI group and 44.6% for the chemotherapy group. At the 24-month cutoff, the 2-year survival rate of 48.2% for the ICI group versus 28.6% for the chemotherapy group.

Fig. 3.

Kaplan-Meier (KM) analysis of overall survival with integrated number at risk.

The median OS was 22 months in the ICI group compared with 11.5 months in the chemotherapy group. Survival distributions differed significantly between the two groups, as determined by the log-rank test (p = 0.009). These findings indicate that ICI monotherapy significantly prolongs overall survival compared with chemotherapy.

Table 3 presents the results of the univariate and multivariate Cox proportional hazards analyses for OS. In the univariate analysis, the treatment group emerged as the only significant predictor of OS. Compared with the ICI group, patients receiving chemotherapy had a significantly increased risk of death (hazard ratio [HR] = 1.836; 95% CI: 1.144–2.948; p = 0.012). No other baseline characteristics were identified as significant predictors in the univariate analysis (all p $>$ 0.05).

For a robust multivariate analysis, treatment group and other clinically relevant variables (age, ECOG performance status, and PD-L1 expression) were included to adjust for potential confounders. After adjustment, the treatment group remained an independent and significant prognostic factor for OS. Chemotherapy was associated with a significantly higher risk of mortality relative to ICI therapy (adjusted hazard ratio [aHR] = 1.796; 95% CI: 1.112–2.900; p = 0.017).

Adverse events (AEs) observed in this study are summarised in Table 4. Clear differences in safety profiles were noted between the two groups, with the most prominent disparities observed in hematologic toxicity. The incidence of leukopenia was significantly higher in the chemotherapy group (p = 0.039). Similarly, anaemia (any grade) was significantly more frequent in the chemotherapy group compared to the ICI group (p = 0.013). irAEs were predominantly observed in the ICI group. Compared with the chemotherapy group, patients receiving ICI therapy experienced significantly higher rates of hypothyroidism (p = 0.027) and rash/pruritus (p = 0.008), both at any grade level.