Safety and effectiveness of nivolumab in patients with locally advanced or metastatic non-small cell lung cancer in China: a nationwide electronic health record dataset analysis

Abstract

Background

While the efficacy and safety of nivolumab have been demonstrated in clinical trials involving Chinese patients with non-small cell lung cancer (NSCLC), real-world evidence on treatment patterns, safety, and clinical outcomes remains limited. This study aimed to investigate the safety profile and effectiveness of nivolumab in Chinese NSCLC patients.

Methods

This retrospective observational cohort included patients diagnosed with advanced NSCLC who received at least one dose of nivolumab between June 1, 2018, and August 31, 2022, as identified from the National Anti-Tumor Drug Surveillance System (NATDSS). Longitudinal electronic health record (EHR) data for the identified patients were used to assess safety outcomes, treatment patterns, overall survival (OS), and real-world progression-free survival (rwPFS). Safety outcomes and treatment patterns were analyzed descriptively. Time-to-event outcomes, including OS and rwPFS, were evaluated using the Kaplan-Meier (KM) method and reported as median time to event with 95% confidence intervals (CIs). Subgroup analyses of safety and effectiveness were conducted according to line of therapy, treatment regimen (monotherapy or combination), and tumor histology.

Results

A total of 2,649 eligible patients with locally advanced or metastatic NSCLC were identified across 224 hospitals in China. The median age was 63.0 years, and the median time from diagnosis to nivolumab initiation was 90.0 days [interquartile range (IQR): 12–327 days]. Nivolumab was administered as first-line therapy in 52.9% (n=1,400), second-line in 31.8% (n=842), and third-line or beyond in 23.1% (n=613) of patients. The median follow-up was 12.9 months (95% CI: 12.4–13.7). Immune-related adverse events (irAEs) occurred in 14.8% (n=391) of the patients, with grade 3–4 irAEs in 2.5% (n=67). The incidence of grade 3–4 irAEs was slightly higher in the combination therapy group than in the monotherapy group (2.5% vs. 1.9%). Serious adverse events (SAEs) were reported in 4.3% (n=114) of patients. No new safety signals were identified in older patients (≥75 years), or those with baseline renal or hepatic impairment. The median OS was 21.2 months (95% CI: 18.7–24.5).

Conclusions

This large-scale, real-world study provides further evidence of manageable safety profile and favorable effectiveness of nivolumab in Chinese patients with advanced NSCLC. These findings support the continued use of nivolumab in appropriate patients and provide reassurance that real-world outcomes are consistent with those observed in clinical trials.

Highlight box.

Key findings

• This is the first nationwide real-world study assessing the safety and effectiveness of nivolumab in Chinese patients with advanced non-small cell lung cancer (NSCLC) using a comprehensive electronic health record (EHR) dataset.

• Nivolumab demonstrated a manageable safety profile and favorable effectiveness in a diverse, real-world Chinese NSCLC population.

• Subgroup analyses showed consistent safety and effectiveness across therapy lines, regimens (monotherapy/combination), histological types, and biomarker-defined subgroups.

What is known and what is new?

• Nivolumab’s efficacy and safety have been established in global and Chinese clinical trials, but real-world evidence in the Chinese population is limited.

• This study provides new large-scale, real-world evidence for nivolumab in Chinese NSCLC patients, including special populations often excluded from clinical trials (elderly, renal/hepatic impairment, first-line use).

What is the implication, and what should change now?

• Results support the continued use of nivolumab in appropriate patients and provide reassurance to clinicians that real-world outcomes align with those observed in controlled studies.

• The study’s data framework and methodology set a new standard for future EHR-based pharmacovigilance and real-world evidence generation in China.

Introduction

Lung cancer remains a formidable challenge in oncology, significantly impacting global health with its high incidence and mortality rates (1). In China, the burden is particularly severe, with over a million new cases and hundreds of thousands of deaths annually (2). Nivolumab, a pioneering anti-programmed cell death protein 1 (PD-1) monoclonal antibody, has revolutionized the treatment landscape for metastatic non-small cell lung cancer (NSCLC) since its approval in the United States, following the pivotal Phase 3 studies, CheckMate 017 (3) and CheckMate 057 (4). This immunotherapy has been embraced for a variety of cancers, including melanoma, squamous cell carcinoma of the head and neck (SCCHN), malignant pleural mesothelioma (MPM), urothelial carcinoma (UC), gastric carcinoma (GC), esophageal squamous cell carcinoma (ESCC), esophageal cancer (EC), esophageal adenocarcinoma (EAC), gastroesophageal junction cancer (GEJC), renal cell carcinoma (RCC), classical Hodgkin lymphoma (cHL), microsatellite instability-high/mismatch repair-deficient (MSI-H/dMMR) colorectal cancer (CRC), and hepatoma carcinoma (HCC), etc.

In China, nivolumab was approved as a second-line therapy for locally advanced or metastatic NSCLC in June 2018, based on the Phase 3 CheckMate 078 study (5). This study demonstrated an increase in median overall survival (mOS) to 11.9 months compared to 9.5 months with docetaxel, with no new safety signals (5). Long-term follow-up revealed a 3-year overall survival (OS) rate of 19% in the nivolumab arm vs. 12% in the docetaxel arm (6). Despite these promising results, the study population in randomized controlled trials (RCTs) often does not fully represent the general patient population, considering factors such as demographics, disease severity, comorbidities, and other characteristics. Moreover, the controlled environment of RCTs may not reflect the variability observed in real-world settings.

Several global observational studies have confirmed the effectiveness of nivolumab, either as monotherapy or in combination, for NSCLC, with no outstanding safety signals (7-10). However, real-world studies focusing on nivolumab in the Chinese population are lacking. Therefore, through a comprehensive analysis of the nationwide oncology electronic health record (EHR) database in China, we aimed to understand the treatment patterns, effectiveness, and safety of nivolumab in Chinese patients. The primary objectives of this study were to evaluate the real-world safety profile, including treatment-emergent adverse events (TEAEs), immune-related adverse events (irAEs), serious adverse events (SAEs), and treatment patterns. The secondary objective was to assess the effectiveness, including OS and real-world progression-free survival (rwPFS) for nivolumab therapies in patients with locally advanced and metastatic NSCLC in China. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1535/rc).

Methods

Study design and data source

This retrospective observational study obtained a nationwide oncology EHR dataset from the National Anti-Tumor Drug Surveillance System (NATDSS) (ClinicalTrials.gov, NCT04825873). Since 2013, NATDSS has collected longitudinal EHR from more than 1,400 hospitals across 31 provinces in China, encompassing more than 15 million cancer patients. The dataset includes 19 categories of variables such as demographics, prescribed drug information, medical data, medical testing, and treatment procedures for both inpatient and outpatient services. Additionally, the dataset is linked to the national death registry from the China Center for Disease Control and Prevention (CDC), ensuring reliable death information, ensuring a reliable death information. The NATDSS EHR dataset has been utilized in other published real-world studies (11-13).

Structured data, including diagnosis, symptoms, laboratory testing, surgery, pathological morphology, and drugs, were harmonized and normalized to a standard ontology based on international standards. Unstructured data were extracted from EHR-based digital documents using regular expression or pre-treated natural language processing models. During quality control process, both manual and automated measures were employed, with over 900 quality control rules addressing relevance, data volume, accuracy, exclusiveness, completeness, normalization, and rationality. To ensure data integrity, study protocol and statistical analysis plan were established prior to execution. Patient confidentiality was protected by de-identifying structured data and implementing measures to prevent re-identification. Data integrity checks were conducted throughout the study, with diligent enforcement and monitoring of data traceability and replicability.

This study was adhered to the International Society for Pharmacoepidemiology (ISPE) Guidelines for Good Epidemiology Practices. Individuals were de-identified. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by ethics committee of The First Affiliated Hospital of Xi’an Jiaotong University (No. XJTU1AF2022LSK-266) and individual consent for this study was waived due to the retrospective nature of the study. The study protocol was approved before initiation.

Study population

The study included all patients with a primary diagnosis of lung cancer identified by International Classification of Diseases (ICD) between June 1, 2018, and August 31, 2022, from NATDSS. Inclusion criteria were locally advanced or metastatic NSCLC treated with at least one dose of nivolumab. Patients with other confirmed concurrent primary tumors or those who received any PD-1/programmed death ligand-1 (PD-L1) inhibitor treatments or other drug targeting immune checkpoint pathways at any time prior to the index date were excluded. The index date was defined as the first day of nivolumab administration.

The cohort for safety assessment and treatment patterns included all eligible patients. The cohort for effectiveness assessment included individuals whose follow-up ended before January 1, 2022, as the CDC death registry was updated through December 31, 2021, at the time of data analysis.

Outcome and key variable definitions

Demographic data included age, sex, body mass index (BMI), and smoking status. The NSCLC stages at the index date included stages III and IV. Comorbidities included lung disease, hepatic disease, autoimmune disease, allergies, hypertension, diabetes, coronary heart disease, and cerebrovascular diseases. Sites of distant metastasis included the lungs, liver, brain, bone, and adrenal gland. Histological types were classified as squamous cell carcinoma, adenocarcinoma, and others. These variables were measured up to 12 months before and up to 2 weeks after the index date. Baseline information was derived from the data closest to the index date if repeated measures were available. The Eastern Cooperative Oncology Group (ECOG) score closest to nivolumab initiation was used, with physician-recorded ECOG scores preferentially extracted. The Karnofsky Performance Status score was converted to the ECOG score (14). When ECOG performance status was not explicitly documented in the medical record, we applied predefined rules to impute values based on available clinical indicators. Specifically, two trained medical staff reviewed physicians’ notes for descriptions of disease symptoms, energy level, functional status, mobility, mode of admission (e.g., transfer, emergency, walking, being assisted, wheelchair, stretcher), self-care ability, bedding time, work capability, and consciousness. These indicators were used to assign a Karnofsky Performance level and derive an ECOG score. An ECOG score was directly assigned if the information is clear. If insufficient information was available, ECOG was coded as missing.

For the safety assessment, TEAEs, irAEs, and SAEs of each patient were collected from the first nivolumab dose to the end of study. Types, severity, and causality were manually extracted by two independent medical research staff members using predefined rules and algorithms. TEAEs included all recorded adverse events following nivolumab administration, manifesting as signs and symptoms, diseases, or abnormal laboratory tests, but were not necessarily causally related to the treatment drug. irAEs were categorized as pulmonary, gastrointestinal, hepatic, endocrine, renal, skin, and other, using predefined medical terms. TEAEs were classified as related, potentially related, or unrelated to nivolumab. Related irAEs were those explicitly described as immune-related by the physician, or TEAEs in patients without prior autoimmune conditions requiring immunosuppressants or dose reduction/discontinuation of nivolumab followed by relief. Potential irAEs were determined by records from physicians. All other TEAEs were considered unrelated. For analysis, related and potentially immune-related TEAEs were combined as immune-related events. AE grading was primarily based on the physician’s description. If unclear, the grading was determined based on the symptoms, severity, and treatment measures described for AE according to Common Terminology Criteria for Adverse Events (CTCAE) 5.0.

Data on treatment patterns, including the line of therapy, treatment regimen, nivolumab usage, and post-nivolumab treatment, were collected. The lines of therapy were defined as follows: first-line treatment was defined as the first anti-tumor therapy the patients received following the diagnosis of locally advanced or metastatic NSCLC. All anticancer medication administered within the first 30 days of the first-line treatment were grouped as first-line regimens. Switching between drugs in the same class was not considered a new regimen. The first administration of a new class of anticancer drugs marked the start of second-line treatment. The nivolumab treatment regimen was divided into two categories: nivolumab monotherapy and nivolumab combination therapy, which included combination therapy with chemotherapy, anti-angiogenic therapy, targeted therapy, and others. Information on received treatment before and after nivolumab treatment was also collected. Prior therapies included radiotherapy, chemoradiotherapy, platinum-based therapy, surgical treatment, neoadjuvant therapy, and adjuvant therapy. Subsequent treatment patterns after nivolumab included chemotherapy, immunotherapy, targeted therapy, and combination therapies.

Clinical effectiveness outcomes included the assessment of OS, OS rate, and rwPFS. OS was defined as the time from the initiation of nivolumab treatment to death from any cause. For patients still alive, OS was censored at the date of last known survival. rwPFS was defined as the time from nivolumab initiation to first tumor progression. Patient progression status was determined based on physician-documented tumor progression (extracted manually), including death. If there was no clear documentation of progression, the progression date was selected as the start date of the next line of therapy following nivolumab treatment, otherwise rwPFS was censored at the date of loss to follow-up after nivolumab treatment.

Statistical analysis

The planned number of patients was 400, which was sufficient to assess the incidence of selected grade 3 or 4 TEAEs. Drawing on data from CheckMate studies 078, 017, and 057, the incidence for these selected grade 3 or 4 TEAEs ranged from 0.3% to 2.1% (3-5). With an analysis involving 400 patients, the study has the capacity to detect the lowest incidence of 0.3%, achieving a 95% confidence interval (CI) of 0.01–1.4% and a statistical power of 80.51%. In this EHR-based retrospective study included all patients who met the predefined inclusion and exclusion criteria in the analytical dataset.

This analysis in this study was primarily descriptive. Categorical variables are reported as counts (n) and frequencies (%). Continuous variables are reported as mean (standard deviation) and median [interquartile range (IQR) or range], depending on the distributions of variables. Time to event outcomes were assessed using the Kaplan-Meier (KM) method and are reported as descriptive statistics (e.g., median time to event) with 95% CIs. Unless otherwise specified, missing data were excluded from percentage calculation.

Subgroup analyses of safety and effectiveness (OS, OS rate, and rwPFS) were performed by lines of therapy, treatment regimen (monotherapy or combination), and tumor histology.

All statistical analyses were performed using SAS statistical software (version 9.4; SAS Institute Inc., Cary, NC, USA).

Results

Patient characteristics

Between June 1, 2018, and December 31, 2022, 2,649 consecutive patients with locally advanced or metastatic NSCLC from 224 hospitals on NATDSS, who received at least one dose of nivolumab, were included in this study (Figure 1).

Figure 1.

Patient flowchart. NSCLC, non-small cell lung cancer; PD-1, programmed death-1; PD-L1, programmed death ligand-1.

Patient demographics and clinical characteristics are summarized in Table 1. The median (range) age was 63.9 years (24–89 years) and 80.1% were male (n=2,121). Among the 1,531 patients with smoking status, 57.6% were current or former smokers. Of the 1,098 patients with ECOG performance data, 10.3% had a score ≥2. Histological types included squamous cell carcinoma (38.6%), adenocarcinoma (44.3%), and other types (17.1%). Among the 240 patients with PD-L1 expression data, 70.8% were PD-L1 positive. At the initiation of nivolumab therapy, 93.1% of the patients had stage IV disease. The most common distant metastatic sites were bone (46.6%), lungs (42.3%), brain (20.7%), and liver (16.1%). Comorbidities included lung disease (50.8%), allergies (36.7%), hepatitis (26.5%), hypertension (21.9%), diabetes (10.8%), cerebrovascular disease (9.3%), coronary disease (5.6%), and autoimmune disease (4.4%).

Table 1. Baseline demographics and clinical characteristics.

Variable	Value
Age, years (n=2,649)	63.9 [24–89]
Age group, years (n=2,649)
<65	1,552 (58.6)
65≤ age <75	901 (34.0)
≥75	196 (7.4)
Sex (n=2,649)
Male	2,121 (80.1)
Female	528 (19.9)
BMI, kg/m2 (n=1,128)	22.8±3.3 (10.8, 33.7)
Smoking status (n=1,531)
Current smoker	561 (36.6)
Former smoker	321 (21.0)
Never smoker	649 (42.4)
ECOG score (n=1,098)
0	222 (20.2)
1	763 (69.5)
≥2	113 (10.3)
Staging with nivolumab (n=2,649)
Stage III	184 (6.9)
Stage IV	2,465 (93.1)
Histological type (n=1,710)
Squamous cell carcinoma	661 (38.6)
Adenocarcinoma	757 (44.3)
Other	292 (17.1)
Site of distant metastasis (n=2,289)
Lung	967 (42.3)
Liver	368 (16.1)
Brain	474 (20.7)
Bone	1,067 (46.6)
Adrenal gland	167 (7.3)
Other sites	557 (24.3)
Comorbidity (n=2,649)
Lung disease	1,345 (50.8)
Hepatic disease	701 (26.5)
Autoimmune disease	116 (4.4)
Allergy	972 (36.7)
Hypertension	579 (21.9)
Diabetes	285 (10.8)
Coronary heart disease	148 (5.6)
Cerebrovascular disease	245 (9.3)
Line of therapy with nivolumab (n=2,649)†
First-line	1,400 (52.9)
Second-line	842 (31.8)
Third-line and higher	613 (23.1)
Treatment regimen of nivolumab (n=2,649)
Monotherapy	825 (31.1)
Combination therapy	1,824 (68.9)
PD-L1 expression (n=240)
Positive	170 (70.8)
Negative	70 (29.2)
Prior therapy (n=2,649)
Radiotherapy	1,135 (42.9)
Chemoradiotherapy	1,020 (38.5)
Platinum-based therapy	836 (31.6)
Surgical treatment	239 (9.0)
Neoadjuvant therapy	49 (1.9)
Adjuvant therapy	179 (6.8)

Data are presented as number (%), median [range], or mean ± standard deviation (min, max). ^†, some patients received nivolumab across multiple therapy lines, resulting in a cumulative treatment total greater than 100%. BMI, body mass index; ECOG, Eastern Cooperative Oncology Group; PD-L1, programmed death ligand-1.

The median follow-up was 12.9 months (95% CI: 12.4–13.7). The median time from diagnosis of locally advanced or metastatic NSCLC to initiation of nivolumab therapy was 90 days (range: 12–327 days). Nivolumab-containing regimens were administered to 1,400 patients (52.9%) as first-line therapy, 842 patients (31.8%) as second-line therapy, and 613 (23.1%) as third-line or higher therapy. Nivolumab monotherapy and combination therapy were used in 23.4% and 76.6% of patients, respectively.

Safety

Among the 2,649 patients, TEAEs occurred in more than half of the patients (53.9%), and the majority (92.6%) of the events were mild or moderate (Table S1). The most common TEAEs included gastrointestinal disorders (9.7%), lung disorders (9.5%), and skin disorders (7.5%). The median time from nivolumab initiation to TEAEs onset was 36 days (range: 0–1,179 days) and the median time from TEAEs onset to resolution was 10 days (range: 0–653 days) (Table S2).

SAEs occurred in 4.3% (114/2,649) of patients, with an incidence rate of 0.07 per person-year of exposure (Table S3). The median time from the first nivolumab dose to SAE onset was 39 days (range: 0–729 days), and the median time from SAE onset to resolution was 12 days (range: 0–205 days) (Table S2). At the end of study, most SAEs were either fully resolved or resolving, while 7.0% (n=8) of patients with an SAE had fatal outcomes (Table S4).

irAEs occurred in 14.8% (n=391) of patients, with 753 events and an overall incidence rate of 0.34 per person-year (Figure 2). Grade 3 or 4 irAEs were less frequent, with 76 events (incidence rate of 0.03 per person-year) reported in 67 patients (2.5%). Among the 391 patients who experienced irAEs, 226 patients had sufficient data to evaluate the time from onset to resolution. The median time from the index date to irAE onset was 44 days (range: 0–791 days), and the median time from irAE onset to resolution was 15 days (range: 0–653 days). By the end of study observation, 52.2% of the irAEs (n=364 events) were resolved or resolving (Table S2).

Figure 2.

irAEs by category and the median time to onset and resolution. The length of each bar represents the time range, and the median is shown as a circle; 0.00 means <0.01 and two decimal were retained when 0.01≤ value <1. ^a, incidence rate per person-year = number of events/total [(last survival time − first dose of nivolumab + 1)/365.25. ^b, resolution referred to patients whose adverse event outcomes were fully recovered or recovering. ^c, the analysis of resolution included irAEs with sufficient data to evaluate the time from onset to resolution. ^d, index date was defined as day 0. ^e, time-to-resolution referred to the time from the onset of an adverse event to the first full recovery or improvement in the corresponding symptoms. ^f, fatal cases were reported, including three pulmonary, one renal, and one other adverse event. irAE, immune-related adverse event.

Within the selected irAE categories, the most common irAE was pulmonary toxicity, affecting 100 patients (3.8%) with 108 events and an incidence rate of 0.05 per person-year. Grade 3–4 pulmonary irAEs only occurred in 0.5% of patients (n=12), with an incidence rate of 0.01 per person-year (Figure 2). Skin disorders occurred in 67 patients (2.5%), with 73 events and an incidence rate of 0.05 per person-year. Endocrine disorders occurred in 58 patients (2.2%) with 66 events and an incidence rate of 0.03 per person-year. irAEs affecting less than 2% of patients included liver function abnormalities, gastrointestinal disorders, and renal disorders (Figure 2).

Among all patients with irAEs, 104 (26.7%) received immunosuppression supportive care, and treatment was permanently discontinued in 29 (7.4%) of patients with irAEs. Only 5 patients (0.7%) had a fatal outcome, which included three pulmonary, one renal, and one other (fever) irAE, reflecting the relative infrequency of life-threatening irAEs with nivolumab (Table S5).

In addition, among 698 patients who received nivolumab monotherapy, 79 patients (11.3%) experienced at least one irAE, and 13 patients (1.9%) experienced grade 3 or 4 events (Table S6). irAEs occurred slightly more frequently in nivolumab combination group (14.9%). Pulmonary toxicity was the most common irAE in both monotherapy and combination therapy (2.72% and 3.41%, respectively). Details are presented in Table S6.

irAEs in specific populations, including elderly patients (≥75 years old) and those with renal or hepatic impairment, were also analyzed (Table S7). There were 196 patients aged ≥75 years, 115 with renal impairment and 269 with hepatic impairment at baseline. The proportion of irAEs (all grades) was nearly the same as that in the overall population: 14.3% (28/196) of elderly patients with 0.33 events per person-year, 13.0% (15/115) of patients with renal impairment (0.65 events per person-year), and 11.9% (32/269) of patients with hepatic impairment (0.69 events per person-year), respectively. Grade 3–4 irAEs occurred infrequently, with lung toxicities being the most common irAEs observed across specific populations.

Treatment patterns

Nivolumab treatment patterns are summarized in Figures 3,4 and Table S8. Nivolumab was administered as first-line therapy in 1,400 patients (52.8%), of whom 401 (28.6%) were treated with nivolumab alone and 999 (71.4%) with nivolumab-containing combination therapy, accounting for 15.1% and 37.7% of all patients, respectively. Chemotherapy (73.7%) was the most common drug combined with nivolumab. Chemotherapy and nivolumab-containing combined regimens were commonly prescribed for patients with subsequent second-line treatment data.

Figure 3.

Treatment patterns of nivolumab in real-world practice settings.

Figure 4.

Different treatment regimens across line of therapy in patients who received nivolumab combination therapy. AI, anti-angiogenesis therapy; CT, chemotherapy; Nivo, nivolumab.

Nivolumab was administered as second-line therapy in 842 patients, accounting for 31.8% of the study population. Among them, 357 patients (42.4%) were treated with nivolumab alone, and 485 (57.6%) with nivolumab-containing combination therapy (combined with chemotherapy in 52% of patients) (Figure 4). Chemotherapy was the most prior line therapy in patients treated with second-line nivolumab. For those with subsequent therapy information, nivolumab-containing combination therapy was the most frequently used (Figure 3).

As third-line therapy or higher, nivolumab-containing combination therapy was the most administered treatment, with 546 patients, accounting for 89.1% of third-line nivolumab and 62.4% of all patients receiving third line and above treatment (Figure 3). In these later lines of treatment, nivolumab combination therapy consisted of various regimens, including anti-angiogenic therapy, chemotherapy, targeted therapy, and dual immunotherapy (Figure 4).

Effectiveness in overall and subgroup populations

The effectiveness analysis dataset included 2,445 patients. Median OS was 21.2 months (95% CI: 18.7–24.5 months), with 6-month, 1-year, 2-year, and 3-year survival rates of 76.1%, 62.3%, 47.1%, and 45.5%, respectively (Figure 5A). The median rwPFS was 7.4 months (95% CI: 6.9–8.1), with 6-month, 1-year, 2-year, and 3-year rwPFS rates of 55.5%, 37.7%, 22.4%, and 18.8%, respectively (Figure 5B).

Figure 5.

Effectiveness of overall population and sub-population by line of therapy. Kaplan-Meier curves for OS and rwPFS in overall population (A,B) (n=2,445), and in patients treated with nivolumab monotherapy (C,D) and combination therapy (E,F) stratified by treatment lines (1st, 2nd, and ≥3rd). All plots are anchored on the initiation of nivolumab treatment. CI, confidence interval; OS, overall survival; rwPFS, real-world progression-free survival.

Effectiveness by line of therapy

A total of 825 patients received nivolumab monotherapy across different lines of therapy. For the 378 patients who received nivolumab as first-line therapy, the median OS was 22.5 months [95% CI: 16.0–not reached (NR)] and the median rwPFS was 9.9 months (95% CI: 8.1–12.7). For the 335 patients who received nivolumab as second-line therapy, the median OS was 19.4 months (95% CI: 14.9–25.9) and the median rwPFS was 4.9 months (95% CI: 4.1–6.0). For the 60 patients who received nivolumab as third-line or higher therapy, the median OS was 8.2 months (95% CI: 3.7–14.3) and the median rwPFS was 4.9 months (95% CI: 2.3–9.4) (Figure 5C,5D).

A total of 1,824 patients received nivolumab as combination therapy across all therapy lines. Among these patients, 864 received nivolumab as first-line therapy, with the median OS not reached and the median rwPFS of 11.5 months (95% CI: 10.0–13.5). For the 464 patients who received nivolumab as second-line therapy, the median OS was 16.1 months (95% CI: 12.9–24.8), and the median rwPFS was 5.7 months (95% CI: 4.9–6.4). For the 530 patients who received nivolumab as third-line therapy or higher therapy, the median OS was 18.6 months (95% CI: 16.7–23.1) and the median rwPFS was 12.4 months (95% CI: 10.1–16.8) (Figure 5E,5F).

Effectiveness by histological types

For the 618 patients with squamous cell carcinoma, the mOS was 22.5 months (95% CI: 17.8–NR), 1-year survival rate was 64.2%, and the median rwPFS was 8.3 months (95% CI: 7.1–9.4) (Figure 6A,6B). For the 718 patients with adenocarcinoma, the median OS was 22.1 months (95% CI: 15.7–NR), 1-year survival rate was 59.9%, and the median rwPFS was 6.2 months (95% CI: 5.5–7.6). For the 266 patients with other histological types, the median OS was 15.6 months (95% CI: 11.0–21.0), 1-year survival rate was 53.2%, and the median rwPFS was 6.7 months (95% CI: 5.6–9.3). The clinical outcomes of different therapy lines under each histology are shown in Figure 6C-6F.

Figure 6.

Clinical outcomes by histology types and stratified by lines of therapy. Kaplan-Meier curves of OS and rwPFS in NSCLC patients by different histology types (A,B), and in patients with squamous NSCLC (C,D), and adenocarcinoma NSCLC (E,F) stratified by the treatment lines in squamous NSCLC patients. All plots are anchored on the initiation of nivolumab treatment. Median number of both OS and rwPFS, 3-year survival or rwPFS rate with 95% CI were shown for each histology type. CI, confidence interval; NSCLC, non-small cell lung cancer; OS, overall survival; rwPFS, real-world progression-free survival.

Effectiveness by PD-L1 expression

For the 202 patients with available PD-L1 expression data at baseline, the median OS for the 143 patients with positive PD-L1 expression was not reached and the median rwPFS was 7.4 months (95% CI: 6.7–14.3). For the remaining patients with negative PD-L1 expression, the median OS was 23.9 months (95% CI: 10.2–NR) and the median rwPFS was 6.7 months (95% CI: 4.2–12.5) (Figure S1).

Discussion

To date, this study represents the largest real-world analysis of the safety and effectiveness of nivolumab in patients with locally advanced or metastatic NSCLC in China. Leveraging a nationwide oncology EHR database encompassing 2,862,241 patients with NSCLC, we identified 2,649 eligible patients treated with nivolumab across 224 hospitals. The study population was heterogeneous, including patients with ECOG scores ≥2, brain metastases, and comorbidities involving vital organs. This diversity enhances the generalizability of our findings and provide a comprehensive reflection of nivolumab use in routine clinical practice in China.

Our results align with those of the CheckMate-078 trial, supporting the manageable safety and favorable effectiveness of nivolumab in a broader NSCLC population. The incidence of overall irAEs and organ-specific irAEs in our study was lower than those reported in CheckMate-078 and other real-world studies (5-10). The observed incidence rates (0.2–0.5 per person-year) were consistent with rates previously published real-world data (7,15). Grade 3–4 irAEs were infrequent (2.5%), falling within the 2–10% range reported in other real-world studies (9,15-18). The relatively low incidence of grade 3–4 irAEs observed in our study may be attributed to several factors. First, improved risk identification and patient selection have enabled clinicians to avoid immune checkpoint inhibitors in individuals at higher risk for severe toxicities. Second, increased clinical experience has led to earlier recognition and more effective management of irAEs, reducing the occurrence of severe cases. Finally, underreporting of adverse events in real-world settings, and patients who receive ongoing care at local hospitals which were not covered by our study database, may have contributed to an underestimation of the true incidence. Pulmonary irAEs were the most common, consistent with the findings of a Japanese post-marketing surveillance study (19). Although prior studies suggested that age and hepatic or renal comorbidities may increase AE risk, our analysis did not find these factors to be associated with higher rates of grade ≥3 irAEs. Additionally, the incidence of irAEs was similar between monotherapy and combination therapy groups.

In terms of effectiveness outcomes, our study demonstrated longer survival compared to the CheckMate-078 and other PD-1/PD-L1 inhibitor trials (20-24). For patients receiving nivolumab monotherapy as second-line treatment, the median OS of patients was 19.4 months and the median rwPFS was 4.9 months compared to 12.0 and 2.8 months, respectively, in CheckMate-078 (5). Compared to other real-world studies of nivolumab reporting median OS of 8–16 months and PFS of 2–7.8 months (7,9,15,17,18,25,26), our findings suggest comparable or superior outcomes.

Notably, nivolumab was frequently used as first-line therapy in clinical practice in China, reflecting evolving treatment patterns and physician preference. While the CheckMate 026 trial did not demonstrate superiority of nivolumab monotherapy over chemotherapy in PD-L1 selected patients as first-line treatment, subsequent studies have expanded the evidence base for immunotherapy in NSCLC (27). For monotherapy in the first-line setting, pembrolizumab has demonstrated significant improvements over chemotherapy alone in PD-L1-selected [tumor proportion score (TPS) >1] populations (median OS 16.7 vs. 12.1 months; median PFS 5.4 vs. 6.5 months) (20,28). In our study, first-line nivolumab monotherapy yielded a median OS of 22.5 months and a median rwPFS of 9.9 months, regardless of PD-L1 expression. When combined with chemotherapy, CheckMate 227 Part 2 reported a numerically improved OS with nivolumab plus chemotherapy compared to chemotherapy alone, regardless of tumor PD-L1 expression (median OS 18.3 vs. 14.7 months) (29). This combination approach was more frequently adopted compared with nivolumab monotherapy (999 vs. 401 patients) in our study, which is consistent with strategies employed for other PD-1/PD-L1 inhibitors. Similar benefits have been observed with camrelizumab in advanced squamous NSCLC (CAMEL trial) and sugemalimab in metastatic NSCLC (GEMSTONE-302 trial), both showing improved survival outcomes when combined with chemotherapy (23,30). Additionally, the recent approval of the nivolumab-ipilimumab combination in China for first-line treatment of metastatic NSCLC with PD-L1 TPS ≥1% and without EGFR/ALK alterations (based on CheckMate 227 Part 1) marks a significant advancement. As the first dual-immunotherapy regimen available in China, its introduction is expected to further diversify and influence first-line treatment strategies (31,32). Regarding the observation that OS in third-line or later settings was longer than in second line, we acknowledge that this may be due to misclassification of treatment lines, which is a recognized limitation in retrospective EHR-based studies. These findings highlight both the dynamic nature of NSCLC management in China and the importance of ongoing real-world evidence to inform clinical decision-making.

Subgroup analyses by histology and PD-L1 expression were consistent with prior immunotherapy studies. Our findings also provide insight into real-world treatment patterns in China, where nivolumab is used across therapy lines, with a preference for combination regimens. These combinations often include chemotherapy or anti-angiogenic agents, reflecting a cautious yet exploratory approach among Chinese clinicians and a growing trend toward using nivolumab in first-line settings.

The major strength of this study lies in its scale and representativeness. It is the largest real-world analysis of nivolumab in Chinese NSCLC patients, utilizing high-quality EHR data with rigorous quality control. The inclusion of a diverse patient population enhances the external validity of our findings. However, limitations inherent to retrospective real-world studies remain. Mild TEAEs and irAEs may be underreported, and rwPFS may be overestimated due to irregular and shorter follow-up and undocumented progression events. Additionally, limited biomarker data, such as PD-L1 expression, restricts deeper exploration of treatment response predictors. Furthermore, the median follow-up time was relatively short due to the study’s data cutoff, which may increase the risk of early censoring and impact the interpretation of survival outcomes. This limitation is compounded by the administrative censoring resulting from missed visits due to patient transitions between treatment centers and local hospitals, leading to incomplete follow-up documentation in the study database.

Conclusions

This large-scale, retrospective real-world study provides further evidence that nivolumab is effective and well-tolerated in Chinese patients with locally advanced or metastatic NSCLC. Subgroup analyses, including irAEs in special populations and effectiveness across treatment lines, regimens, and biomarker-defined subgroups, provide valuable insights to inform clinical decision-making in routine practice.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by ethics committee of The First Affiliated Hospital of Xi’an Jiaotong University (No. XJTU1AF2022LSK-266) and individual consent for this study was waived due to the retrospective nature of the study.

Data Sharing Statement

Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1535/dss

jtd-18-02-90-dss.pdf ^{(67.7KB, pdf)}

DOI: 10.21037/jtd-2025-1535